[House Hearing, 112 Congress]
[From the U.S. Government Printing Office]
CLOUD COMPUTING: WHAT ARE THE SECURITY IMPLICATIONS?
=======================================================================
HEARING
before the
SUBCOMMITTEE ON CYBERSECURITY,
INFRASTRUCTURE PROTECTION,
AND SECURITY TECHNOLOGIES
of the
COMMITTEE ON HOMELAND SECURITY
HOUSE OF REPRESENTATIVES
ONE HUNDRED TWELFTH CONGRESS
FIRST SESSION
__________
OCTOBER 6, 2011
__________
Serial No. 112-50
__________
Printed for the use of the Committee on Homeland Security
[GRAPHIC(S) NOT AVAILABLE IN TIFF FORMAT]
Available via the World Wide Web: http://www.gpo.gov/fdsys/
__________
U.S. GOVERNMENT PRINTING OFFICE
73-737 PDF WASHINGTON : 2012
-----------------------------------------------------------------------
For sale by the Superintendent of Documents, U.S. Government Printing
Office Internet: bookstore.gpo.gov Phone: toll free (866) 512-1800; DC
area (202) 512-1800 Fax: (202) 512-2104 Mail: Stop IDCC, Washington, DC
20402-0001
COMMITTEE ON HOMELAND SECURITY
Peter T. King, New York, Chairman
Lamar Smith, Texas Bennie G. Thompson, Mississippi
Daniel E. Lungren, California Loretta Sanchez, California
Mike Rogers, Alabama Sheila Jackson Lee, Texas
Michael T. McCaul, Texas Henry Cuellar, Texas
Gus M. Bilirakis, Florida Yvette D. Clarke, New York
Paul C. Broun, Georgia Laura Richardson, California
Candice S. Miller, Michigan Danny K. Davis, Illinois
Tim Walberg, Michigan Brian Higgins, New York
Chip Cravaack, Minnesota Jackie Speier, California
Joe Walsh, Illinois Cedric L. Richmond, Louisiana
Patrick Meehan, Pennsylvania Hansen Clarke, Michigan
Ben Quayle, Arizona William R. Keating, Massachusetts
Scott Rigell, Virginia Kathleen C. Hochul, New York
Billy Long, Missouri Janice Hahn, California
Jeff Duncan, South Carolina
Tom Marino, Pennsylvania
Blake Farenthold, Texas
Robert L. Turner, New York
Michael J. Russell, Staff Director/Chief Counsel
Kerry Ann Watkins, Senior Policy Director
Michael S. Twinchek, Chief Clerk
I. Lanier Avant, Minority Staff Director
------
SUBCOMMITTEE ON CYBERSECURITY, INFRASTRUCTURE PROTECTION, AND SECURITY
TECHNOLOGIES
Daniel E. Lungren, California, Chairman
Michael T. McCaul, Texas Yvette D. Clarke, New York
Tim Walberg, Michigan, Vice Chair Laura Richardson, California
Patrick Meehan, Pennsylvania Cedric L. Richmond, Louisiana
Billy Long, Missouri William R. Keating, Massachusetts
Tom Marino, Pennsylvania Bennie G. Thompson, Mississippi
Peter T. King, New York (Ex (Ex Officio)
Officio)
Coley C. O'Brien, Staff Director
Alan Carroll, Subcommittee Clerk
Chris Schepis, Minority Senior Professional Staff Member
C O N T E N T S
----------
Page
STATEMENTS
The Honorable Daniel E. Lungren, a Representative in Congress
From the State of California, and Chairman, Subcommittee on
Cybersecurity, Infrastructure Protection, and Security
Technologies................................................... 1
The Honorable Bennie G. Thompson, a Representative in Congress
From the State of Mississippi, and Ranking Member, Committee on
Homeland Security.............................................. 3
WITNESSES
Panel I
Mr. Richard Spires, Chief Information Officer, U.S. Department of
Homeland Security:
Oral Statement................................................. 5
Prepared Statement............................................. 6
Mr. David McClure, Ph.D., Associate Administrator, Office of
Citizen Services and Innovative Technologies, U.S. General
Services Administration:
Oral Statement................................................. 12
Prepared Statement............................................. 14
Mr. Gregory C. Wilshusen, Director of Information Security
Issues, Government Accountability Office:
Oral Statement................................................. 18
Prepared Statement............................................. 19
Panel II
Mr. James W. Sheaffer, President, North American Public Sector,
Computer Sciences Corporation:
Oral Statement................................................. 38
Prepared Statement............................................. 40
Mr. Timothy Brown, Senior Vice President and Chief Architect for
Security, CA Technologies:
Oral Statement................................................. 43
Prepared Statement............................................. 45
Mr. James R. Bottum, Vice Provost for Computing and Information
Technology and Chief Information Officer, Clemson University:
Oral Statement................................................. 52
Prepared Statement............................................. 54
Mr. John Curran, Chief Executive Officer, American Registry of
Internet Numbers:
Oral Statement................................................. 62
Prepared Statement............................................. 64
APPENDIX
Questions From Honorable William Keating For Richard Spires...... 73
CLOUD COMPUTING: WHAT ARE THE SECURITY IMPLICATIONS?
----------
Thursday, October 6, 2011
U.S. House of Representatives,
Committee on Homeland Security,
Subcommittee on Cybersecurity, Infrastructure Protection,
and Security Technologies,
Washington, DC.
The subcommittee met, pursuant to call, at 10:02 a.m., in
Room 311, Cannon House Office Building, Hon. Daniel E. Lungren
[Chairman of the subcommittee] presiding.
Present: Representatives Lungren, Walberg, Marino, Clarke,
Richardson, Keating, and Thompson.
Also present: Representative Duncan.
Mr. Lungren. We have been informed that we are probably
going to have votes at 8--I mean, at 10:20, or something, and
then have about four or five votes, and so we will have a delay
for our hearing for about 45 minutes. So we are going to try
and get started very quickly, get our opening statements in and
begin your testimony, and then we will have to break and beg
your indulgence on that.
The Committee on Homeland Security Subcommittee on
Cybersecurity, Infrastructure Protection, and Security
Technologies will come to order. The subcommittee is meeting
today to examine the security implications of cloud computing.
I would recognize myself for an opening statement.
We welcome our witnesses today and look forward to their
testimonies regarding cloud computing phenomena. According to
NIST, cloud computing delivers I.T. services and applications
to users by enabling ubiquitous, convenient, on-demand network
access to a shared pool of configurable computing resources.
Cloud computing enables organizations and individuals to
access website data and on-line programs without concern about
the server's physical location, thereby promising cheaper,
faster, more flexible, more effective information technology.
Most organizations already utilize some form of cloud
computing. On-line shopping and banking are prime examples of
how cloud computing has transformed the way in which companies
interact with and provide on-line services to customers.
Improved technologies over the years have increased our
computing capabilities and reduced costs. This new cloud
technology also promises greater I.T. capability at reduced
cost.
The administration has issued a Cloud First policy to
accelerate the pace at which Government evaluates safe and
secure cloud computing options before making any new I.T.
investment. Republican Members of Congress, and I hope our
Democratic colleagues, are always looking for ways to reduce
Government spending, so any savings from cloud computing would,
indeed, be welcome.
However, in spite of this projected I.T. savings, we cannot
ignore our responsibility as Members of this Cybersecurity
Subcommittee to assure that Government information will be
secure in the cloud. GAO reported last spring that security
incidents at Government agencies rose 650 percent over the last
5 years.
Our concern is the cloud offers--that the cloud offers a
rich target for hackers, criminals, terrorists, and rogue
nations. With cyber-espionage affecting every sector of our
economy, aggregating important information in one location is a
legitimate security concern. You might say it is a target-rich
environment.
Security implications cannot be an afterthought. Obviously,
they need to be considered as cloud technology is being
developed and deployed.
Yesterday we Republicans released our House task force
recommendation for cybersecurity legislation. We intend to work
with our colleagues on the other side of the aisle because this
is not a partisan issue; it is one that we need more work on,
and I do believe there is a bipartisan commitment to provide
that work. Speaker Boehner has made cybersecurity a top
priority, and our committee will be a key player in drafting
House legislation.
So as we address our numerous cyber vulnerabilities we must
scrutinize new technologies and their attendant risks to ensure
that further vulnerabilities will not be created. Cloud
advocates argue that even sensitive data can be secure in the
cloud. They argue that the cloud providers have the resources
to invest at sophisticated security--in sophisticated security
systems if necessary.
Different security levels can be designed for the various
cloud configurations. The private cloud is appropriate for
classifying the most sensitive of personnel data, we are told.
Sensitive data can be--can use the hybrid model; nonsensitive
data can use the public cloud.
While I.T. savings are important, we cannot ignore the
information security risk created by cloud technology.
Assessing those risks responsibly will be critical if cloud
computing is ever going to be widely accepted.
The Federal cloud computing strategy is designed to ensure
the security of Government information and establish a
transparent security environment between cloud service
providers and the Federal Government. NIST and the General
Services Administration have developed the Federal Risk and
Authorization Management Program, FedRAMP, to facilitate and
lead the development of standards for security,
interoperability, and portability.
The strategy states that the transition to a cloud
computing environment is an exercise in risk management that
entails identifying and assessing risk and taking steps to
reduce it to an acceptable level. We look forward to the
testimony of Dr. McClure, from GSA, will outline this important
FedRAMP program.
Today we intend to examine the benefits and risks of cloud
computing, and hopefully identify its security implications. I
look forward to the testimony of all of our witnesses this
morning regarding this new cloud technology.
I would now recognize the Ranking Member of the full
committee, Mr. Thompson, for any statement that he might make.
Mr. Thompson. Thank you very much, Mr. Chairman. Before I
begin my statement, let me take off on your comments about the
Republican caucus' release of its cyber task force
recommendations yesterday.
As you know, cyber is an emerging homeland security threat
that warrants timely bipartisan action from Congress. The
stakes are high and Federal networks alone have seen a 650-fold
increase in cyber attacks over the past 5 years.
As you know, the President has submitted to Congress a
comprehensive plan, including a legislative proposal. Taking
your comments that you look forward to a bipartisan effort on
this issue, I can assure you from our side of this committee,
we will do just that.
With respect to this morning's hearing on security
implications of cloud computing, cloud computing can and does
mean different things to different people. The National
Institute of Standards and Technology, NIST, has published a
definition that provides a starting place for discussing and
defining security needs, but not everyone agrees with or
conforms to NIST's definition. So as of today, the Federal
Government and industry have not reached agreement about how
uniform rules and standards that should be adopted to secure
the information in the cloud.
This is not something that can be left up in the air. While
I embrace technological progress, I also know that every new
technology presents great possibilities as well as great
challenges. In our eagerness to jump on the bandwagon we often
forget to ask about the destination of the wagon, the cost of
the journey, and the roads which we will take along the way.
As we embark on this new journey of migrating information
to the cloud we must not repeat mistakes of the past. We must
be about some of the claims that are made.
For instance, I am told that the cloud will produce cost
savings and create efficiencies. I am told that these benefits
will be achieved by eliminating the need for data centers,
computer hardware, and other public and private sector
operations that employ thousands of people. I have to ask about
these displaced people.
While every new technology creates displacement, it also
provides opportunities. So we must ask what new opportunities
will be provided and who will benefit?
Finally, as cloud computing increases the Federal
Government's ability to communicate effectively, we must ask
how to increase the ability to communicate will affect the
security of Government operations.
Mr. Chairman, without clear standards and uniform rules we
cannot begin to evaluate how the security of Government data
will be affected by cloud computing. Additionally, we must
remember that cloud computing must be aligned with the Federal
Information Security Management Act, FISMA.
Given that the Federal Government currently uses the
services of external vendors to manage its cloud operations, we
must ask how these businesses will comply with FISMA
regulations governing auditing and security requirements.
Industries cannot effectively compete without understanding the
potential regulatory environment that will be caused by
widespread use of cloud computing in the Federal Government.
Mr. Chairman, there are many questions that must be
resolved. However, I am certain that our witnesses today will
be able to shine some light on the cloud.
I yield back.
Mr. Lungren. Thank you very much, Ranking Member, for that
poetic opening statement.
When the Ranking Member of the subcommittee appears we will
give her an opportunity to make her opening statement. Other
Members of the committee are reminded that opening statements
may be submitted for the record.
We are pleased to have a very distinguished panel of
witnesses before us today on this important topic.
Richard Spires was appointed as the chief information
officer of the Department of Homeland Security 2009. He has
extensive knowledge in senior level operations and information
technology issues through working both the public and the
private sectors. Previously oversaw I.T. responsibilities for
the Internal Revenue Service as deputy commissioner for
operations support, chief information officer and associate
information officer for business systems modernization
respectively.
Before joining the IRS he served as the president, chief
operation officer, and director of Mantas, Inc., a software
product vendor. He also spent more than 16 years at SRA
International, a systems integration company.
Welcome.
Dr. David McClure was appointed associate administrator of
the U.S. General Services Administration's Office of Citizen
Services and Innovative Technologies in 2009. Dr. McClure most
recently served as the managing vice president for Gartner,
Inc.'s government research team.
Before working at Gartner, Dr. McClure served as vice
president for e-government and technology at the Council for
Excellence in Government. He has also had an 18-year career
with the Government Accountability Office.
Greg Wilshusen--is that the proper----
Mr. Wilshusen. Perfect.
Mr. Lungren. Thank you--is director of information security
services at the Government Accountability Office. He has spent
over 28 years of auditing, financial management, and
information systems prior to this date.
Prior to joining GAO in 1997, he was the senior systems
analyst at the Department of Education and served as the
comptroller for the North Carolina Department of Environment,
Health, and Natural Resources; and held senior auditing
positions at Irving Burton Associates, Inc. and the U.S. Army
Audit Agency.
Thank you, gentleman, for all being here. We have the rule
of a 5-minute testimony. We have your written statements; they
will be included in their totality in the record. We would ask
you to go in the order in which I introduced you.
So, Mr. Spires, the Chairman would now recognize you.
STATEMENT OF RICHARD SPIRES, CHIEF INFORMATION OFFICER, U.S.
DEPARTMENT OF HOMELAND SECURITY
Mr. Spires. Chairman Lungren, Mr. Thompson, and Members of
the subcommittee, thank you and good morning. Today I will
discuss the changes cloud computing is having within the
Government and at the Department of Homeland Security. Also, I
will discuss how DHS is addressing the security challenges
associated with cloud computing.
Simply, cloud computing enables Federal agencies to
purchase on-demand I.T. services using a consumption-based
business model. Liken cloud computing to the electric power or
telecommunications markets: we, as customers, pay for the usage
of the service itself, whether it be so much per kilowatt-hour
for electric power or minutes of usage for the use of our cell
phone. As I.T. matures, many services are becoming commoditized
and lend themselves to such a usage-based model.
Cloud computing is truly transforming the I.T. business
because it does provide significant benefits to customers. The
cloud provides scalability and rapid deployment, full
transparency for managing operational costs, and controlling
and reducing capital expenses.
Further, cloud computing simplifies the overall
administration and cost of I.T. infrastructure. Early
projections for DHS look to yield cost avoidance savings of 8
to 10 percent once we transition to cloud infrastructure
services.
DHS is taking an aggressive approach to the use of cloud
computing, with 12 DHS offerings either in production, awarded,
or in the acquisition phase. DHS is currently focused on two
deployment models: Our private cloud and the use of the public
cloud.
For the DHS private cloud, we manage sensitive information
within our two enterprise data centers and use our internal
wide-area network. A few examples of DHS private cloud
offerings include our Email as a Service, which we expect to
have more than 100,000 users live by the end of fiscal year
2012.
SharePoint as a Service will support more than 90,000 users
by the end of this calendar year. Development and Test as a
Service provides a development and test environment linked to
the production environment we enable--to enable successful
deployment of new applications. We expect to provision new
servers within 1 business day with this new capability, while
the legacy model averaged up to 6 months.
WorkPlace as a Service will provide secure, virtual desktop
access that seamlessly support mobile devices, to include cell
phones and tablets. This service will better enable a mobile
DHS workforce to support telework and continuity of operations.
We are embracing the use of public cloud services to manage
nonsensitive information. DHS has successfully deployed Self
Check in the public cloud, and over the next 2 years will
consolidate its public-facing websites, like dhs.gov, to the
public cloud.
To effectively manage security risks of cloud computing DHS
is leveraging our private cloud environment to enable services
to manage sensitive information. The model bolsters information
security through our defense-in-depth strategy.
By hosting in the enterprise data centers the DHS private
cloud can leverage the existing enterprise security controls as
well as leverage the use of our continuous monitoring
capabilities and trusted internet connections. By embedding
enhanced enterprise security controls in our private cloud, DHS
will provide security assurance exceeding that of our existing
legacy systems.
For public clouds there is a visibility gap between the
provider and customer in which they cannot see into each
other's management, operational procedures, and technical
infrastructure. To address security concerns of public cloud
offerings, this visibility gap must be reduced through a series
of requirements for contractual reporting and technical
auditing and continuous monitoring data feeds to verify that
the provider and customer are meeting their responsibilities.
The FedRAMP program will help Federal agencies address
these challenges as they leverage public cloud providers or
establish their own private cloud. Continued work on the
information security challenges will increase the defensive
capabilities of cloud offerings, increasing the assurance level
and the ability for Federal agencies to use public cloud
computing for more sensitive information.
Looking ahead 5 years, the cloud service commodity market
appears poised to grow exponentially. Federal CIOs must focus
on preparing departments and agencies to welcome innovation and
changes in the way we do business. Already, at DHS we are
seeing reduced time to market for new capabilities, reducing
our capital expenditures, and gaining transparency into our
operational expenses, all while providing improved service.
The benefits of cloud computing far outweigh the
challenges.
Thank you.
[The prepared statement of Mr. Spires follows:]
Prepared Statement of Richard Spires
October 6, 2011
Chairmen Lungren, Ranking Member Clarke, and Members of the
subcommittee, thank you and good morning. Today, I will discuss the
changes Cloud Computing is having within the Government and industry
and how the Department of Homeland Security (DHS) is pursuing this
capability to enhance mission performance and gain efficiencies in
Information Technology (IT). This testimony also will provide an
overview of the current state of cloud computing at the Department of
Homeland Security, outlining the Department's initiatives to move data
to the cloud in order to implement the White House's ``Cloud First''
policy as specified in the ``Federal Cloud Computing Strategy'' issued
February 8, 2011, and the ``25 Point Implementation Plan to Reform
Federal Information Technology Management'' issued December 9, 2010.
Finally, I will address the IT security challenges associated with
cloud computing and how DHS is addressing such challenges.
moving to the cloud
First, allow me to explain what cloud computing is and why it is so
vital. The legacy IT model of separate IT infrastructures for each
system--both within the Federal Government and industry--must evolve to
meet the growing customer demands within a budget-constrained
environment. The traditional model is not well-positioned to reduce
time to market for new services or provide transparency for operational
expenses. It also introduces higher risk due to up-front capital
expenditures. Additionally, customized applications hosted in
traditional data center environments cannot scale fast enough to
support urgent demand in real-time. These challenges, in addition to
potential security vulnerabilities, present a call to action for the
Federal Government and industry.
Fortunately, we are experiencing an exciting change within the IT
industry--the rise of cloud computing. This evolutionary transformation
is fast replacing the legacy IT model not only within private industry
but also within the Federal Government.
The National Institute of Standards and Technology (NIST), an
agency of the U.S. Department of Commerce, provides the following
definition of cloud computing in NIST Special Publication 800-145 (NIST
SP 800-145):
``Cloud computing is as a model for enabling convenient, on-demand
network access to a shared pool of configurable computing resources
(e.g., networks, servers, storage, applications, and services) that can
be rapidly provisioned and released with minimal management effort or
service provider interaction. This cloud model promotes availability
and is composed of five essential characteristics, three service
models, and four deployment models.''
Cloud computing provides the rapid delivery of computing resources
inexpensively to multiple users from a centralized source of related
and unique service offerings that is shared by many customers. To
provide further context, this model is similar to business models
deployed in the electric power, cable, or telecommunications markets.
That is, within this model, customers do not fund up-front costs to
fully stand up environments, or fund on-going operations and
maintenance costs. Instead these capital costs are borne by industry,
while the customer only pays for services received in the consumption-
based model.
NIST prescribes the following five primary characteristics of cloud
computing:
1. On-demand self-service.--A consumer can unilaterally provision
computing capabilities, such as server time and network
storage, as needed automatically without requiring human
interaction with each service's provider.
2. Broad network access.--Capabilities are available over the
network and accessed through standard mechanisms.
3. Resource pooling.--The provider's computing resources are pooled
to serve multiple consumers using a multi-tenant model, with
different physical and virtual resources dynamically assigned
and reassigned according to consumer demand.
4. Rapid elasticity.--Capabilities can be rapidly and elastically
provisioned, in some cases automatically, to quickly scale out,
and rapidly released to quickly scale in. To the consumer, the
capabilities available for provisioning often appear to be
unlimited and can be purchased in any quantity at any time.
5. Measured Service.--Cloud systems automatically control and
optimize resource use by leveraging a metering capability at
some level of abstraction appropriate to the type of service.
Resource usage can be monitored, controlled, and reported,
providing transparency for both the provider and consumer of
the utilized service.
NIST also identifies three discrete service offerings, each of a
unique value to the customer. As customers move up this offering chain,
they gain greater efficiencies, yet more standardization is required:
1. Cloud Infrastructure as a Service (IaaS).--The capability
provided to the consumer is to provision processing, storage,
networks, and other fundamental computing resources where the
consumer is able to deploy and run arbitrary software, which
can include operating systems and applications. The consumer
does not manage or control the underlying cloud infrastructure
but has control over operating systems, storage, deployed
applications, and possibly limited control of select networking
components (e.g., host firewalls). This model provides the most
flexibility for the customer, however will not provide all the
potential efficiencies gained at the Software as a Service
model.
2. Cloud Platform as a Service (PaaS).--The capability provided to
the consumer is to deploy onto the cloud infrastructure
consumer-created or acquired applications created using
programming languages and tools supported by the provider. The
consumer does not manage or control the underlying cloud
infrastructure including network, servers, operating systems,
or storage, but has control over the deployed applications and
possibly application hosting environment configurations.
3. Cloud Software as a Service (SaaS).--The capability provided to
the consumer is to use the provider's applications running on a
cloud infrastructure. The consumer does not manage or control
the underlying cloud infrastructure including network, servers,
operating systems, storage, or even individual application
capabilities, with the possible exception of limited user-
specific application configuration settings.
Finally, NIST identifies four primary deployment models, which are
generally accepted across Government. These deployment models range
from models that are more secure to those that are more available.
Federal agencies will employ models based on risk-based decisions that
address their financial, operational, and security needs. The four
models include:
1. Private cloud.--The cloud infrastructure is operated solely for
an organization. It may be managed by the organization or a
third party and may exist on-premise or off-premise.
2. Community cloud.--The cloud infrastructure is shared by several
organizations and supports a specific community that has shared
concerns (e.g., mission, security requirements, policy, and
compliance considerations). It may be managed by the
organizations or a third party and may exist on-premise or off-
premise.
3. Public cloud.--The cloud infrastructure is made available to the
general public or a large industry group and is owned by an
organization selling cloud services.
Hybrid cloud.--The cloud infrastructure is a composition of two or
more clouds (private, community, or public) that remain unique
entities but are bound together by standardized or proprietary
technology that enables data and application portability (e.g.,
cloud bursting for load-balancing between clouds).
DHS is currently focused on two of the four deployment models,
private cloud and public cloud. DHS will house our private cloud
computing capabilities within our two enterprise data centers, while
our public cloud will be hosted by organizations selling cloud
services. I will provide more detail on these momentarily, but first
allow me to briefly address the differences between the cloud and the
traditional IT business model.
the benefits and risks of cloud computing
Cloud computing is truly transforming the IT business. It is
difficult to say which is more compelling--the cloud's significant
scalability and rapid deployment, or full transparency for managing
operational costs. For many, controlling and reducing capital expense
(the expenditures used to acquire physical assets, including both
equipment and office space) is uppermost, while others argue meeting
demand is the foremost concern. The cloud addresses both and is clearly
becoming vital to how we align IT to support mission and business
requirements.
For example, the deployment of private cloud services at DHS
enables the Department's many components to outsource hosting and other
services capabilities to DHS's two Enterprise Data Centers (EDCs). This
model enables components to pay on a per-use basis, rather than
standing up isolated capabilities throughout the organization that
duplicate efforts and costs. In fact, early projections for these
services look to yield cost avoidance savings of 8 to 10 percent once
we fully transition to private cloud infrastructure services.
As DHS moves more of its operations to cloud computing models, it
will simplify the overall administration and oversight of its IT
infrastructure. DHS will move from having to manage operations of its
infrastructure at the server level, to one in which DHS ensures that
cloud-based service level agreements (SLAs) are being met by the
service provider. Such simplification will enable discretionary
resources to be moved to better understanding and fulfilling customer
needs, so that IT organizations can focus more of their efforts on
addressing core business and mission needs.
Migration to the cloud, however, is not without information
security risks. The Federal Cloud Computing Strategy specifies:
. . . it is not sufficient to consider only the potential value of
moving to cloud services. Agencies should make risk-based decisions
which carefully consider the readiness of commercial or government
providers to fulfill their Federal needs.
It is important to recognize that many Federal departments and
agencies are targeted by Advanced Persistent Threat (APT) campaigns by
adversaries that attempt to compromise Government information systems
to further their own objectives. These APT campaigns are aggressive,
well-financed, and difficult to detect and prevent. APTs target the
systems necessary to achieve their goals, regardless of the cloud or
traditional computing environments in use by the Federal department or
agency. Some cloud environments have capabilities necessary to defend
against and provide recovery from these threats, such as advanced
monitoring capabilities and cleared information security professionals,
while other cloud environments may not, because the increased costs to
provide these security capabilities may price their cloud offering
outside of the competitive marketplace for their customers. Thus, the
security capabilities of the cloud offering must be considered to
determine cloud readiness before use by a Federal department or agency,
and why DHS considers use of both public and private cloud computing
important, as I will discuss later.
building the cloud at dhs
At DHS, we are pursuing private and public cloud offerings.
Specifically, we are establishing private cloud services to manage
sensitive but unclassified information, while using the public cloud
for non-sensitive information. We have already made significant strides
through nine DHS cloud service offerings that are either in the
planning, acquisition, or sustainment phase.
DHS has committed to nine current and planned private cloud
services:
Email as a Service (EaaS).--DHS is in the process of rolling
out our messaging capability across Headquarters and Federal
Emergency Management Agency (FEMA). We expect to have more than
100,000 users DHS-wide on this service offering by the end of
fiscal year 2012.
SharePoint as a Service (SHPTaaS).--We are currently
migrating Headquarters and United States Citizenship and
Immigration Services (USCIS) users to our secure collaboration
program. We expect to have nearly 90,000 users DHS-wide on this
service by the end of the 2011 calendar year. This migration
will significantly improve information-sharing capabilities
across DHS.
Development and Test as a Service (DTaaS).--Establishing
development and test offerings in the cloud will have
tremendous positive impact on DHS. Currently, DHS has multiple
development environments spread across the Department and
industry locations. Because all environments are different,
moving new releases to production or changes to existing
environments presents high-risk and multiple challenges and new
releases or changes may not always work in production, leading
to significant inefficiencies. Moving and hosting development
and test services to our enterprise data centers provides not
only a simple path to transition from project creation to
implementation, but also accelerated delivery. In fact, we
expect to provision new servers within 1 business day with this
new capability, while the legacy model averaged up to 6 months
to provision one server. Additionally, this service will
provide on-demand testing and application management tools,
which will significantly improve the quality of our new
offerings. DHS plans to roll out DTaaS over the next 60 days.
Infrastructure as a Service (IaaS).--Complementary to the
Development and Test as a Service (DTaaS) offering is our
Infrastructure as a Service (IaaS) offering to provide
virtualized production services, including operating systems,
network, and storage, that is consistent with new industry
standards. These services will provide a logical destination
for code developed in the development and test environment. We
aim to stand up new services in the cloud in less than 1 week,
while the legacy model typically averaged up to 12 to 18
months. DHS expects to have initial IaaS capabilities by the
end of the 2011 calendar year.
WorkPlace as a Service (WPaaS).--Enabling a mobile workforce
is a priority within the Department. We are working closely
with the Department's other line-of-business chiefs to
modernize how DHS employees work. This offering will provide
robust virtual desktop, remote access, and other mobile
services over the next 24 months. This capability enables
telework and Continuity of Operations (COOP), not only in the
National Capital Region (NCR), but for DHS personnel Nation-
wide. Additionally, we expect to reduce our out-year
expenditures on traditional desktop and laptops as we consume
more mobile enabling technologies.
Project Server as a Service (PSaaS).--This offering will
provide a robust project management platform to publish project
schedules that can more easily be shared across offices,
divisions, and components. We expect this service to better
enable standardization of project management disciplines and
directly support our efforts to improve the management of both
IT and non-IT programs. DHS plans to make available PSaaS
service within the next 30 days.
Authentication as a Service (AuthaaS).--We have already
established a core fundamental offering that provides robust
authentication services across 250,000 Federal and contractor
employees. This service eliminated the need for duplicative
authentication services, while significantly enhancing the
Department's information-sharing needs. Nearly 70 DHS
applications are using this service today.
Case and Relationship Management as a Service (CRMaaS).--
Over the next 6 months, we will rollout our Case and
Relationship Management offering. This offering, leveraging
Enterprise License Agreements (ELAs), will better enable CRM
and case workflows across DHS. Utilizing these services, the
Department will be piloting a litigation case management
capability for ICE, partnering with TSA on modernizing the
redress service, improving customer relationship capabilities
within USCIS, and deploying a regulations tracking service for
DHS.
Business Intelligence as a Service (BIaaS).--The Department
is already piloting an early version of a Business Intelligence
capability which started in March 2011 and will run through
fiscal year 2012. The Department will leverage this current
offering to enhance transparency into departmental programming
and expenditures. By the end of fiscal year 2012, we expect the
Department will have visibility to information sources across
the investment life-cycle, including IT, financial, human
resources, asset management, and other information sources.
Based on the successful pilot and maturing offerings in
service, the Department will look to move to a full Business
Intelligence as a Service offering in fiscal year 2013.
Establishing these private cloud services is critical to our
success. Our private cloud offerings will provide real value to the
organization. As mentioned previously, private cloud services will
enable components to outsource secure, commodity IT services to DHS's
two enterprise data centers to eliminate redundancy and reduce costs,
while ensuring information security. Each service will be rolled out
with a minimum ``Federal Information Security Management Act of 2002''
(FISMA) rating of Moderate or High. Clearly, our private cloud services
will streamline our time to market and enhance our security posture,
better enabling DHS to accomplish its mission.
But DHS is not wedded to only establishing private cloud services
at its two enterprise data centers. We are embarking on a public cloud
strategy as well. The Department will leverage public cloud
capabilities to enhance Government-to-citizen-services and gain
operational and financial efficiencies. In addition, the FedRAMP
initiative will address critical security concerns of agency Chief
Information Officers (CIOs) over the next few years by having cloud
services receive provisional security authorities to operate.
The Department has three public cloud initiatives underway. Two are
already deployed, and the third will be piloting in Quarter 1 of fiscal
year 2012.
Identity Proofing as a Service (IDPaaS).--We successfully
deployed an innovative identity proofing service in the cloud
in March 2011. This offering met USCIS's EVerify Self Check
requirement to allow individuals in the United States to check
their employment eligibility status before formally seeking
employment and is the first on-line E-Verify program offered
directly to workers and job seekers. This service is now
available in more than 20 States, including the District of
Columbia. This voluntary, free, fast, and secure service was
developed through a partnership between the DHS and the Social
Security Administration (SSA).
Enterprise Content Delivery as a Service ECDaaS.--For the
past several years, DHS has used cloud service for Enterprise
Content Delivery (ECD) to ensure our public-facing websites are
always available. The private sector uses this capability
extensively, and DHS adopted EDC for protection against denial
of service attacks, to help manage surge requirements, and to
significantly reduce hosting costs. This service proved
invaluable during the July 4, 2009, denial of service attack on
multiple Federal websites. DHS.gov experienced a nearly 100-
fold increase in traffic, and no services were lost to the
public. The Department has 70% of its externally-facing
websites using this service today.
Web Content Management as a Service (WCMaaS).--Finally,
building off our success with our ``RestoretheGulf.gov''
implementation in the public cloud in late fiscal year 2010,
the Department awarded a public cloud hosting contract off the
General Services Administration's (GSA) Infrastructure as a
Service (IaaS) Blanket Purchase Agreement (BPA). Within this
offering, the Department will leverage open source software
hosted in the public cloud and consolidate all public-facing
DHS websites. We expect to complete this consolidation over the
next 2 years. During the next 6 months, the Department will
pilot multiple websites in the cloud, including websites from
U.S. Immigration and Customs Enforcement (ICE), United States
Citizenship and Immigration Services (USCIS), and Federal the
Emergency Management Agency (FEMA).
DHS has taken an aggressive stance regarding the use of both
private and public cloud computing services. The Department continues
to evaluate its enterprise needs, and we certainly expect to deploy
additional cloud services. Further, as the FedRamp model is deployed
across the Federal Government, we anticipate that there will be a
number of public cloud offerings that have been provisionally certified
at the FISMA Low and Moderate levels within the next 2 years. Given
DHS's mission, we believe a robust private cloud solution will always
be needed for DHS's most sensitive applications and data. Further
leverage of public cloud services will enable the Government to ensure
there is robust competition for such services, driving down costs and
improving overall service levels.
securing the cloud at dhs
As stated earlier, at DHS, we are pursuing private and public cloud
offerings, and the DHS cloud security strategy employs both public and
private cloud services as a risk mitigation tool. The move to DHS's
private cloud model bolsters information security through the DHS IT
security Defense-in-Depth (DiD) strategy. DiD is built upon a robust
security architecture and enterprise architecture, and adopts the NIST
definition of private cloud computing. Hosting in the enterprise data
centers is a primary feature of the DHS private cloud and provides
multiple subordinated services, allowing components and systems to
inherit the inherent enterprise security controls for system security.
The DHS private cloud includes the full DHS enterprise security
capabilities outlined in the DiD, including security operations,
OneNet, Trusted Internet Connections (TICs), and Policy Enforcement
Points (PEPs). The technologies are from the various programs within
the layers of the DiD and aids in combating advanced threats, providing
enterprise security controls to all users in DHS, regardless of their
component and mission function.
For the DHS private cloud, we are leveraging continuous monitoring
and migration to common controls at the DHS data centers. Embracing
information security controls through an inherited approach allows
large, complex organizations like DHS to build on economies of scale in
a private cloud infrastructure to reduce the workload for individual
system owners. As common controls are defined and vetted by the DHS
enterprise and provided as a service to system owners, only the system-
specific controls need to be defined and implemented by system owners.
By centrally managing the development, implementation, and assessment
of enterprise common security controls at the DHS enterprise data
centers and through the DHS private cloud, security responsibility can
be shared across multiple information systems.
While private clouds incorporate new technologies that may be
challenging to secure, public clouds introduce additional risks that
must be addressed through controls and contract provisions that ensure
appropriate accountability and visibility. Though many distinctions can
be drawn between public and private cloud computing, a fundamental
measure of readiness is their ability to meet security requirements. By
design, FedRAMP provides a common security risk model that supplies a
consistent baseline for cloud-based services, including security
accreditation designed to vet providers and services for reuse across
Government. Reducing risk and bolstering the security of clouds, while
ensuring the delivery of the promised benefits, FedRAMP not only
applies to public cloud services, but private, too. Ultimately the
consumption of cloud services requires acknowledgement of a shared
responsibility and governance. From the fact that accountability can
never be outsourced from the Authorizing Official (AO) to the need to
continue to meet Government requirements, all require acknowledgement
of a shared responsibility between the cloud service provider and
customer. For public clouds, there is a ``visibility gap'' between the
provider and customer, in which they cannot see into each other's
management, operational, and technical infrastructure, and procedures.
As such, the visibility gap must be reduced through a series of
requirements for contractual reporting and technical auditing and
continuous monitoring data feeds. The key to secure use of cloud
computing is the shared understanding of the division of security
responsibilities between provider and client, and the ability to verify
that both are meeting their responsibilities. As DHS advances in the
use of public cloud computing, we will be ensuring we have the proper
visibility based on a determination of risk given the cloud service and
underlying data in order to ensure the security of our information.
new challenges for cios
While cloud computing is fundamentally changing Federal Government
IT, it is not without its challenges. The decision to embrace cloud
computing is a risk-based management decision, supported by inputs from
stakeholders, including the CIO, Chief Information Security Officer
(CISO), Office of General Counsel (OGC), privacy official, and the
program owners. From a security perspective, agency CIOs face a number
of issues in delivering both private and public cloud capabilities.
These issues range from determining different levels of security
visibility and responsibilities, ensuring strong authentication,
adopting and implementing standards for cloud portability and
interoperability, to establishing contingency planning that recognizes
cloud computing is a shared capability and identifying new
opportunities for real-time continuous monitoring capabilities but
require new audit technologies implemented within the cloud
environment.
Cloud computing also leads to significant management and governance
shifts for a department or agency. CIOs must work closely with
acquisition, procurement, and finance communities to address the new
business paradigm represented by cloud computing. While cloud computing
requires some technological change, the most significant changes will
be to the business and contracting models. Such models will need to
ensure that agencies can move forward effectively with cloud solutions
while maintaining necessary Federal control and oversight, complying
with Federal procurement and competition laws and requirements, and
managing funding limitations. CIOs must also address changes to the
workforce based on this changing paradigm. As the cloud transforms the
way CIOs deliver IT service, the traditional roles of IT specialists
change, too. CIOs must provide leadership to update skills for existing
personnel and recruit new staff in an environment under significant
change.
These challenges are already inherent in the CIO's role. And, they
have one thing in common--change. Perhaps above all, the cloud
challenges CIOs to lead cultural change within their organization.
the future of the cloud
Looking forward, as FedRAMP and Federal acquisition models mature,
the options for Federal agencies to leverage public and community
clouds clearly provide real value to citizens. Continued work on
information security challenges will increase the defensive
capabilities of cloud offerings, increasing the assurance level and the
ability for Federal agencies to use cloud computing for more sensitive
information.
For example, community clouds could provide agencies with a suite
of specialized cloud hosting services that include the standard IaaS,
PaaS, and SaaS offerings with a more robust security, business, and
mission portfolio offerings such as financials, law enforcement,
intelligence, medical/health, and the increased security and privacy
controls necessary to process more sensitive information. The value of
a community of cloud offerings across a broad suite of verticals for
customers may be realized as the true evolution of the cloud in the
years to come.
Looking 5 years into the future, the cloud service commodity market
appears poised to grow exponentially, creating significant innovation
as a result of intense competition. Federal CIOs must focus on
preparing departments and agencies to help foster and welcome
innovation that changes the way we do business. By embracing the
opportunities of cloud computing, we will redefine the role and
capabilities of IT in the Federal Government.
While we in the Federal Government face challenges to successfully
implementing cloud capabilities to enhance mission performance and
realize cost efficiencies, the benefits far outweigh the challenges.
Already at DHS we are seeing reduced time to market for new
capabilities, and soon, we will begin to reduce our capital
expenditures while gaining transparency into our operational
expenditures in ways we have never been able to before. In conclusion,
we should not think of the cloud as simply a technology opportunity. It
is a far more interesting discourse--and a significant change to the
fundamental business model for how IT is delivered in the Federal
Government.
Thank you.
Mr. Lungren. Thank you very much.
Dr. McClure.
STATEMENT OF DAVID MC CLURE, PH.D., ASSOCIATE ADMINISTRATOR,
OFFICE OF CITIZEN SERVICES AND INNOVATIVE TECHNOLOGIES, U.S.
GENERAL SERVICES ADMINISTRATION
Mr. McClure. Thanks, Mr. Chairman.
Good morning, Mr. Thompson and Mr. Keating.
Thanks for having me here on behalf of GSA to talk about
cloud computing and cloud security.
I just wanted to start by making two critical points about
cloud computing itself. It really offers a compelling
opportunity to substantially improve the efficiency of the
Federal Government. When it is implemented with sound security
risk management approaches, cloud computing can ensure more
consistent protection of the Government's I.T. infrastructure,
our data, and our applications.
Second, the practical use of cloud computing really offers
substantial performance benefits for Government. For example,
tangible cost reductions resulting from more efficient data
storage, web hosting, and even analytics performed on our vast
data repositories.
It can enhance productivity by shifting some of our
workforce to high-value process improvement activity, problem
solving, and customer service excellence. It allows us greater
flexibility and scalability, as Richard just talked about--the
ability of CIOs to actually stand-up services in hours, days,
rather than months, and in some cases, years. It allows or
creates an improved self-service environment: On-line,
streamlined, commodity-like purchasing for I.T. resources
rather than a very long and arduous I.T. acquisition.
We are playing a leadership role in facilitating easy
access to cloud-based solutions from commercial providers that
meet Federal requirements, such as virtualization technologies
for our data centers in the Government, cloud e-mail, disaster
recovery and backup, and infrastructure storage. Our
Government-wide procurement vehicles enable agencies to
evaluate viable cloud computing options that meet their
business needs.
Now let me turn to cloud security. Cloud computing, like
any technology, presents both known and new risks alongside the
benefits that it offers. Different types of cloud services--
public, private, community, hybrid--create their own set of
security challenges in the Government setting.
To address these risks in a more uniform and comprehensive
manner we will soon launch a new Government-wide cloud security
program.
Mr. Chairman, you referred to it, the FedRAMP program.
We have worked in close collaboration with cybersecurity
and cloud experts in NIST, DOD, DHS, NSA, OMB, the Federal CIO
Council, and with private industry. Let me be real clear: The
intent of FedRAMP is to strengthen existing security practices
associated with cloud computing solutions, which, in turn, will
build greater trust between providers and consumers and
accelerate appropriate adoption of security cloud solutions
across the Government.
FedRAMP ensures consistency and quality of system security
certification and accreditation; it creates a transparent and
trusted security environment in Government that will incentive
more reusability of security testing and authorizations; and it
fosters the push toward near real-time security assistance
monitoring. It does this by standing up six critical
capabilities.
It standardizes a minimal baseline for Government-wide
security controls for low and moderate risk cloud systems based
upon existing NIST standards and additional controls vetted
with all interested parties. It manages a process for
accrediting independent third-party assessors to ensure greater
competency, consistency, and compliance with required
Government security controls.
It creates a joint authorization board, comprised of CIOs
and technical representatives from DOD, DHS, and GSA, to grant
provisional authorizations for cloud systems that can be
leveraged by multiple agencies. It also allows agencies to
focus on their own specific security requirements and address
legitimate deltas with the baseline controls rather than
repeating work already competently done by another Federal
entity.
Consistent with FISMA changes, it requires cloud service
providers to perform continuous monitoring, especially for
persistent threats, and will eventually automate the exchange
of status information on specific controls on a near-time--near
real-time basis. In concert with DHS, it controls and manages
the incident response, mitigation, and proof of resolution for
FedRAMP-authorized cloud systems.
Last, it will create a secure data repository to facilitate
Government access to security authorization packages, sample
contract language and templates, examples of cloud service-
level agreements, best practices, and continuous monitoring
information.
So, Mr. Chairman, we think these kinds of steps can really
advance more secure cloud computing in the Government. I am
happy to answer questions for the subcommittee.
[The prepared statement of Mr. McClure follows:]
Prepared Statement of David McClure
October 6, 2011
Chairman King, Ranking Member Thompson, and Members of the
subcommittee: Thank you for the opportunity to appear before you today
to discuss the General Service Administration's (GSA) leadership role
in on-going efforts to enable and accelerate adoption of secure cloud
computing across the Federal Government. Cloud adoption is a critical
component of the administration's plan to improve management of the
Government's IT resources. The IT reforms we have underway are enabling
agencies to use information more efficiently and effectively,
delivering improved mission results at lower cost.
cloud computing adoption in the federal government
Before I discuss the security of cloud computing, and the Federal
Risk Authorization and Management Program (FedRAMP) in particular, I
would like to make a two important points. First, cloud computing
offers a compelling opportunity to substantially improve the efficiency
of the Federal Government. It moves us from buying and managing
physical assets to purchasing IT as a commoditized service. Agencies
pay for only IT resources they use in response to fluctuating program
demands, avoiding the expenses of building and maintaining costly IT
infrastructure. When implemented with sound security risk management
approaches, cloud computing also ensures more consistent protection of
the Government's IT infrastructure, data, and applications.
Second, practical use of cloud computing offers substantial
performance benefits for the Government. Federal agencies are moving to
consolidate and virtualize the more than 2,000 Federal data centers.
Cloud technologies provide an ideal path forward to maximize value in
IT investment dollars while substantially lowering costs--an essential
focus given Federal budget constraints. Case studies we have collected
from agencies point to benefits that include:
``tangible cost reductions (data storage, web hosting and analytics
performed on the Government's vast data repositories);
``enhanced productivity (shifting workforce to more high-value process
improvements, problem solving, and customer service excellence);
``greater flexibility and scalability (enabling CIOs to be much more
responsive to pressing service delivery expectations); and
``improved self-service capabilities (on-line streamlined commodity-
like purchasing for IT resources rather than long, arduous IT
acquisitions).''
GSA is playing a leadership role in facilitating easy access to
cloud-based solutions from commercial providers that meet Federal
requirements. This will enable agencies to analyze viable cloud
computing options that meet their business and technology modernization
needs, while reducing barriers to safe and secure cloud computing. We
are developing new cloud computing procurement options with proven
solutions that leverage the Government's buying power. These cloud
procurement vehicles ensure effective cloud security and standards are
in place to lower risk and foster Government-wide use of cloud
computing solutions such as virtualization technologies for Government
data centers, cloud e-mail, disaster recovery/backup, and
infrastructure storage. Useful information about cloud computing and
available solutions is accessible from our web page, Info.Apps.gov.
GSA's Federal Cloud Computing Initiative was started and is managed
under GSA's e-Government program. In fiscal year 2010 and fiscal year
2011 GSA's Federal Cloud Computing Initiative (FCCI) Program Management
Office (PMO) focused on five primary tasks:
Establishing procurement vehicles that allow agencies to
purchase IT resources as commodities, culminating in the award
of the Infrastructure as a Service (IaaS) Blanket Purchase
Agreement under GSA Schedule 70 to 12 diverse cloud service
providers;
Addressing security risks in deploying Government
information in a cloud environment--resulting in the
development of the Federal Risk Authorization Management
Program (FedRAMP);
Establishing a procurement vehicle that will allow agencies
to purchase cloud-based e-mail services, which created GSA's
Email as a Service (EaaS) Blanket Purchase Agreement;
Supporting the Government-wide collection and assessment of
data center inventories, and assisting agencies in the
preparation and execution of plans to close and consolidate
data centers. Current work includes developing a comprehensive
data center Total Cost Model for agencies to use to analyze
alternative consolidation scenarios, enables data-driven
decision-making for infrastructure cost and performance
optimization. Operationalizing a data center marketplace that
would help optimize infrastructure utilization across
Government by matching agencies with excess computing capacity
with those that have immediate requirements is also being
pursued.
Creating apps.gov, an on-line storefront that provides
access to over 3,000 cloud-based products and services where
agencies can research solutions, compare prices and place on-
line orders using GSA's eBuy system.
Initial funding provided by the e-Gov Fund has allowed GSA to be an
effective catalyst for secure cloud technology adoption Government-
wide. However, there are critical activities that still need to be
accomplished to fully realize the significant cost savings and
productivity improvements that GSA can help agencies achieve. The
continuation of these cost-saving initiatives is dependent on fiscal
year 2012 eGov Fund budget levels and decisions.
fedramp: ensuring secure cloud systems adoption
Cloud computing--like any technology--presents both known and new
risks alongside the many benefits outlined above. To address these
risks in a more uniform and comprehensive manner, we will soon launch a
new Government-wide cloud security program--the Federal Risk and
Authorization Management Program (FedRAMP). The primary goal of the
administration's Cloud First policy is to achieve widespread practical
use of secure cloud computing to improve operational efficiency and
effectiveness of Government. Today, each agency typically conducts its
own security Certification and Accreditation (C&A) process for every IT
system it acquires, leading to unnecessary expense, duplication, and
inconsistencies in the application of NIST-derived security controls
testing, evaluation, and certification procedures. According to the
2009 FISMA report to Congress, agencies reported spending $300 million
annually on C&A activities alone.
At GSA, we have worked in close collaboration with cybersecurity
and cloud experts in NIST, DHS, DoD, NSA, OMB, and the Federal CIO
Council and its Information Security and Identity Management
Subcommittee (ISIMC) to develop FedRAMP. An OMB policy memo officially
establishing the FedRAMP program is expected shortly. The intent is to
strengthen existing security practices associated with cloud computing
solutions which, in turn, will build greater trust between providers
and consumers and accelerate appropriate adoption of secure cloud
solutions across Government. Accordingly, FedRAMP establishes a common
set of baseline security assessment and continuous monitoring
requirements for FISMA low- and moderate-impact risk levels using NIST
standards that must be adhered to by all cloud systems. Figure 1
illustrates how FedRAMP will address three fundamental challenges with
how the Federal Government approaches ensuring cloud security.
[GRAPHIC(S)] [NOT AVAILABLE IN TIFF FORMAT]
Ensuring Consistency and Quality in Cloud Security Certification and
Accreditation
FedRAMP approves qualified, independent, third-party security
assessment organizations, ensuring consistent assessment and
accreditation of cloud solutions based on NIST's long-standing
conformity assessment approach. As noted above, security C&As are
currently performed with varying quality and consistency. This is true
for situations where a third-party service provider is contracted to do
a security assessment of a CSP-provided system, product, or service and
where Government security organizations perform the work themselves. As
a result, trust levels are low for reusing this work across agencies.
To address this challenge, FedRAMP will require that cloud services
providers be assessed using these approved, independent, third-party
assessment organizations (3PAOs). The 3PAOs will initially apply for
accreditation through the FedRAMP PMO and be assessed using established
conformity assessment criteria developed by NIST. This will ensure
higher-quality assessments, done much more consistently, using agreed-
upon FedRAMP security assessment controls. This can save millions of
dollars in expenses borne both by Government and industry in running
duplicative assessments of similar solutions by each agency.
Building Trust and Re-Use of Existing C&A Work
All IT systems, including cloud solutions, must receive an
Authority to Operate (ATO) from the buying agency before they can be
made available for purchase and implemented. The ATO is based on a
thorough review by agency security professionals of the security
packages submitted following the C&A process described above. To
accelerate cloud adoption and enable C&A re-use, FedRAMP will provide a
single, provisional authorization that can be used by all agencies as
the basis for issuing an ATO. If additional security assessment
evaluation and testing is needed for specific agency cloud
implementations, the C&A should only address any additional controls
needed above the existing FedRAMP-approved baseline.
FedRAMP establishes a Joint Authorization Board (JAB) that reviews
all cloud systems that have been assessed by approved 3PAOs using
FedRAMP controls and processes. The JAB membership consists of CIOs and
Technical Representatives from DOD, DHS, and GSA. The JAB reviews the
C&A work and decides whether to grant the ``provisional
authorization''--a seal of approval on the C&A work. The security
packages, assessments and documented decisions will be accessible
within Government from a secure central repository. While each agency
must grant its own ATO for systems under its control, FedRAMP will
facilitate greater use of an ``approve once, and use often'' approach,
leveraging more ATOs across Government.
Moving Towards More Real-Time Security Assurance
FedRAMP shifts risk management from annual reporting under FISMA to
more robust continuous monitoring, providing real-time detection and
mitigation of persistent vulnerabilities and security incidents. Using
the expertise of industry, NIST, NSA, DHS, and ISIMC, nine initial
continuous monitoring controls have been identified that are among the
most common persistent threat vulnerabilities in cloud and non-cloud
systems environments. Cloud Service Providers (CSPs) must agree to
near-real time reporting of continuous monitoring data feeds to DHS
and/or agency Security Operations Centers (SOCs). We are finalizing
data reporting details, with the expectation that the process will
eventually use automated data feeds to maximize efficiencies and
timeliness. When done in addition to the C&A evaluations, this will
result in valuable situational cyber awareness--a relevant and timely
picture of a CSP's security posture. In addition, this approach
provides visibility of prompt mitigation and tangible evidence of
resolution; ensuring quick steps are taken to minimize threats to
Government data and operations.
In short, FedRAMP offers the following improvements for cloud
security assessments conducted in the Federal Government:
[GRAPHIC(S)] [NOT AVAILABLE IN TIFF FORMAT]
There is strong support and demand for stronger cloud security from
agencies seeking to adopt cloud services, as required by the
administration's Cloud First policy. Industry cloud services providers
need to know the specific cloud security capabilities for which they
are accountable. They also desire more efficiency in how C&As and ATOs
are leveraged Government-wide to avoid unnecessary, duplicative, costly
security evaluations. Ensuring IT security is an on-going challenge. We
fully expect to make improvements to the process based on collaboration
with all key stakeholders, including industry, lessons learned, and the
continuous evolution of security standards and controls based upon the
careful, deliberative work of NIST.
FedRAMP will be launched in phases that incrementally build toward
sustainable operations and allows for risk management by capturing on-
going lessons learned and process improvement. Initial rollout will
occur this Fall. Initial Operational Capabilities will have limited
scope and cover a relatively small number of cloud service providers.
Full operations are expected to begin next Spring with more robust
operational capabilities and larger intake of cloud service providers
for FedRAMP review and approval. Late in 2012, we expect sustaining
operations to scale by demand using a privatized board for 3PAO
accreditation. We will discuss the rollout in more depth with the
Congress, Government executive branch agencies, industry, and the
public prior to the initial launch date.
conclusion
Considerable progress has been made in adopting successful cloud
solutions. ``Cloud computing'' is now an accepted part of the Federal
IT lexicon. However, there continues to be a need for more thorough
understanding of cloud deployment models, unique security implications,
and data management challenges. Agency executives should not focus on
cloud technology itself; rather, they should focus on the desired
outcome driving the need for cloud adoption delivered in a secure
environment.
FedRAMP will provide a sound, cost-effective framework for secure
cloud computing. CIOs need to work with their line of business
executives and program managers to develop and deploy effective cloud
roadmaps that address pressing agency mission needs, taking into
account appropriate security and risk management. Agencies should
analyze business needs and identify cloud solutions that best fit their
requirements by making secure cloud adoption part of an overall IT
portfolio management and sourcing strategy. Consistent with the Federal
Cloud Computing Strategy, NIST is currently working on the first draft
of a USG Cloud Computing Technology Roadmap, to be released for public
comment in November, 2011. If linked to cloud provider products and
services, it would greatly assist in this decision-making.
Mr. Chairman, thank you for the opportunity to appear today. I look
forward to answering questions from you and Members of the
subcommittee.
Mr. Lungren. Thank you very much, Dr. McClure.
Now, Mr. Wilshusen.
STATEMENT OF GREGORY C. WILSHUSEN, DIRECTOR OF INFORMATION
SECURITY ISSUES, GOVERNMENT ACCOUNTABILITY OFFICE
Mr. Wilshusen. Chairman Lungren, Mr. Thompson, Mr. Keating,
thank you for the opportunity to participate in today's hearing
on cloud computing security. I believe this is a vitally
important topic.
Earlier this week GAO issued a report on Federal
information security in which we note that the number of
security incidents reported by Federal agencies increased by
over 650 percent during the past 5 years. This fact helps to
underscore the need for effective security in cloud computing
environments.
Today I will describe the information security implications
of Federal use of cloud computing services. I will also discuss
GAO's previous reporting on Federal efforts and guidance on
cloud computing and agencies' actions to implement our
recommendations to improve cloud security.
But if I may, Mr. Chairman, I have first like to recognize
Assistant Director Vijay D'Souza and Shaunyce Wallace, from my
staff, who are here, and also Nancy Glover, who is not here,
for their diligent efforts in reviewing cloud security as well
as preparing my statement.
Mr. Chairman, cloud computing can have both positive and
negative information security implications. Potential security
benefits include those related to broad network access,
possible economies of scale, and the use of self-service
technologies. For example, Federal agencies frequently cited
the prospect of on-demand security controls, the consistent
application of those controls, and low-cost disaster recovery
and data storage as potential benefits.
However, the use of cloud computing can also create
numerous information security risks. Twenty-two of the 24 major
Federal agencies reported that they were either concerned or
very concerned about the potential security risks with cloud
computing.
These risks include the ineffective or noncompliant
security practices of the service provider, an inability to
examine controls of the provider, the prospect of data leakage
to unauthorized users, and the loss of data if the cloud
service is terminated. These risks generally relate to
dependence on the security practices and assurances of the
service provider and the sharing of computing resources.
In a report GAO issued last year, we noted that Federal
agencies had begun efforts to address information security for
cloud computing, but specific guidance was lacking and efforts
remained incomplete. We also reported that OMB and GAO--I am
sorry, GSA--had launched Government-wide initiatives but had
not completed key actions pertaining to cloud computing
security.
For example, OMB had not finished its cloud computing
strategy or defined how information security issues would be
addressed in that strategy. Accordingly, in that report GAO
made recommendations to OMB, GSA, and NIST to take several
actions to address these issues.
Since that report was issued in May 2010 these agencies
have made progress in implementing our recommendations, but
additional actions are still needed to assist agencies in
securely implementing cloud computing. For example, in February
OMB issued its cloud computing strategy, which does reference
the establishment of FedRAMP and other security issues;
however, it does not address the need for agency-specific
guidance, the use of standards for control assessments of cloud
service providers, or the division of security responsibilities
between customer and provider.
Consistent with our recommendation, GSA, in collaboration
with the CIO Council, further developed FedRAMP, as Mr. McClure
has indicated in his opening remarks, and intends to issue
additional guidance on FedRAMP later this quarter. In addition,
NIST has issued three of four guidance documents related to
cloud computing and expect to finalize guidelines on security
and privacy in the public cloud computing later this quarter.
These actions and the issuance of appropriate guidance will
help, yet the true test will be their effective implementation
over time.
To summarize, Mr. Chairman, the use of cloud computing
offers the promise of efficient service, but it also carries
risk. OMB, GSA, and NIST have taken steps to develop a
strategy, processes, and guidance on cloud computing security.
Nevertheless, continued efforts will be needed to ensure that
cloud computing is implemented securely in the Federal
Government.
Mr. Chairman, this concludes my statement. Be happy to
answer any questions.
[The prepared statement of Mr. Wilshusen follows:]
Prepared Statement of Gregory C. Wilshusen
October 6, 2011
information security: additional guidance needed to address cloud
computing concerns
Chairman Lungren, Ranking Member Clarke, and Members of the
subcommittee: Thank you for the opportunity to participate in today's
hearing on the security implications of cloud computing. My statement
today summarizes our report issued last year, titled Information
Security: Federal Guidance Needed to Address Control Issues with
Implementing Cloud Computing \1\ and describes actions taken by Federal
agencies to implement our report's recommendations.
---------------------------------------------------------------------------
\1\ GAO, Information Security: Federal Guidance Needed to Address
Control Issues with Implementing Cloud Computing, GAO-10-513
(Washington, DC: May 27, 2010).
---------------------------------------------------------------------------
Cloud computing, an emerging form of delivering computing services,
can, at a high level, be described as a form of computing where users
have access to scalable, on-demand information technology (IT)
capabilities that are provided through internet-based technologies.
Examples of cloud computing include web-based e-mail applications and
common business applications that are accessed on-line through a
browser, instead of through a local computer. Cloud computing can
potentially deliver several benefits over current systems, including
faster deployment of computing resources, a decreased need to buy
hardware or to build data centers, and more robust collaboration
capabilities. However, along with these benefits are the potential
risks that any new form of computing services can bring, including
information security breaches, infrastructure failure, and loss of
data. Media reports have described security breaches of cloud
infrastructure and reports by others have identified security as the
major concern hindering Federal agencies from adopting cloud computing
services.
My statement today will provide a description of: (1) The
information security implications of using cloud computing services in
the Federal Government, (2) our previous reporting on Federal efforts
and guidance to address cloud computing information security, and (3)
our recommendations and subsequent actions taken by Federal agencies to
address Federal cloud computing security issues. In preparing this
statement, we summarized the content of our May 2010 report on cloud
computing security. In conducting the work for that report, we
collected and analyzed information from industry groups, private sector
organizations, the National Institute of Standards and Technology
(NIST), and 24 major Federal agencies.\2\ In addition, we followed up
with agencies to determine the extent to which the recommendations made
in that report have been implemented. The work for the report on which
this statement is based was performed in accordance with generally
accepted Government auditing standards.
---------------------------------------------------------------------------
\2\ The 24 major Federal agencies are the Agency for International
Development; the Departments of Agriculture, Commerce, Defense,
Education, Energy, Health and Human Services, Homeland Security,
Housing and Urban Development, the Interior, Justice, Labor, State,
Transportation, the Treasury, and Veterans Affairs; the Environmental
Protection Agency; the General Services Administration; the National
Aeronautics and Space Administration; the National Science Foundation;
the Nuclear Regulatory Commission; the Office of Personnel Management;
the Small Business Administration; and the Social Security
Administration.
---------------------------------------------------------------------------
background
We have previously reported that cyber threats to Federal
information systems and cyber-based critical infrastructures are
evolving and growing.\3\ Without proper safeguards, computer systems
are vulnerable to individuals and groups with malicious intentions who
can intrude and use their access to obtain and manipulate sensitive
information, commit fraud, disrupt operations, or launch attacks
against other computer systems and networks.
---------------------------------------------------------------------------
\3\ GAO, Cybersecurity: Continued Attention Needed to Protect Our
Nation's Critical Infrastructure and Federal Information Systems, GAO-
11-463T (Washington, DC: Mar. 16, 2011) and Cybersecurity: Continued
Attention Needed to Protect Our Nation's Critical Infrastructure, GAO-
11-865T (Washington, DC: July 26, 2011).
---------------------------------------------------------------------------
In addition, the increasing interconnectivity among information
systems, the internet, and other infrastructure presents increasing
opportunities for attacks. For example, since 2010, several media
reports described incidents that affected cloud service providers such
as Amazon, Google, and Microsoft. Additional media reports have
described hackers exploiting cloud services for malicious purposes. The
adoption of cloud computing will require Federal agencies to implement
new protocols and technologies and interconnect diverse networks and
systems while mitigating and responding to threats.
Our previous reports and those by agency inspectors general
describe serious and widespread information security control
deficiencies that continue to place Federal assets at risk of
inadvertent or deliberate misuse, mission-critical information at risk
of unauthorized modification or destruction, sensitive information at
risk of inappropriate disclosure, and critical operations at risk of
disruption. We have also reported that weaknesses in information
security policies and practices at major Federal agencies continue to
place confidentiality, integrity, and availability of sensitive
information and information systems at risk. Accordingly, we have
designated information security as a Government-wide high-risk area
since 1997,\4\ a designation that remains in force today.\5\ To assist
agencies, GAO and agency inspectors general have made hundreds of
recommendations to agencies for actions necessary to resolve control
deficiencies and information security program shortfalls.
---------------------------------------------------------------------------
\4\ GAO, High-Risk Series: Information Management and Technology,
GAO/HR-97-9 (Washington, DC: February 1997).
\5\ GAO, High-Risk Series: An Update, GAO-11-278 (Washington, DC:
February 2011).
---------------------------------------------------------------------------
Cloud Computing Is a Form of Shared Computing with Several Service and
Deployment Models
Cloud computing delivers IT services by taking advantage of several
broad evolutionary trends in IT, including the use of
virtualization.\6\ According to NIST, cloud computing is a means ``for
enabling convenient, on-demand network access to a shared pool of
configurable computing resources that can be rapidly provisioned and
released with minimal management effort or service provider
interaction.'' NIST also states that an application should possess five
essential characteristics to be considered cloud computing: On-demand
self-service, broad network access, resource pooling, rapid elasticity,
and measured service.
---------------------------------------------------------------------------
\6\ Virtualization is a technology that allows multiple software-
based virtual machines with different operating systems to run in
isolation, side-by-side on the same physical machine. Virtual machines
can be stored as files, making it possible to save a virtual machine
and move it from one physical server to another.
---------------------------------------------------------------------------
Cloud computing offers three service models: Infrastructure as a
service, where a vendor offers various infrastructure components;
platform as a service, where a vendor offers a ready-to-use platform on
which customers can build applications; and software as a service,
which provides a self-contained operating environment used to deliver a
complete application such as web-based e-mail. Figure 1 illustrates
each service model.
[GRAPHIC(S)] [NOT AVAILABLE IN TIFF FORMAT]
In addition, four deployment models for providing cloud services
have been developed: Private, community, public, and hybrid cloud. In a
private cloud, the service is set up specifically for one organization,
although there may be multiple customers within that organization and
the cloud may exist on or off the premises. In a community cloud, the
service is set up for related organizations that have similar
requirements. A public cloud is available to any paying customer and is
owned and operated by the service provider. A hybrid cloud is a
composite of the deployment models. Figure 2 further illustrates each
model.
[GRAPHIC(S)] [NOT AVAILABLE IN TIFF FORMAT]
cloud computing has both positive and negative information security
implications
Cloud computing can both increase and decrease the security of
information systems. Potential information security benefits include
the use of virtualization and automation to expedite the implementation
of secure configurations for virtual machine images. Other advantages
relate to cloud computing's broad network access and use of internet-
based technologies. For example, several agencies stated that cloud
computing provides a reduced need to carry data in removable media
because of the ability to access the data through the internet,
regardless of location. In response to the survey we conducted for our
2010 report, 22 of the 24 major agencies also identified low-cost
disaster recovery and data storage as a potential benefit.
The use of cloud computing can also create numerous information
security risks for Federal agencies. In response to our survey, 22 of
24 major agencies reported that they are either concerned or very
concerned about the potential information security risks associated
with cloud computing. Several of these risks relate to being dependent
on a vendor's security assurances and practices. Specifically, several
agencies stated concerns about:
the possibility that ineffective or non-compliant service
provider security controls could lead to vulnerabilities
affecting the confidentiality, integrity, and availability of
agency information;
the potential loss of governance and physical control over
agency data and information when an agency cedes control to the
provider for the performance of certain security controls and
practices; and:
potentially inadequate background security investigations
for service provider employees that could lead to an increased
risk of wrongful activities by malicious insiders.
Of particular concern was dependency on a vendor. All 24 agencies
specifically noted concern about the possibility of loss of data if a
cloud computing provider stopped offering its services to the agency.
For example, the provider and the customer may not have agreed on terms
to transfer or duplicate the data.
Multitenancy, or the sharing of computing resources by different
organizations, can also increase risk. Twenty-three of 24 major
agencies identified multitenancy as a potential information security
risk because, under this type of arrangement, one customer could
intentionally or unintentionally gain access to another customer's
data, causing a release of sensitive information. Agencies also stated
concerns related to exchanging authentication information on users and
responding to security incidents. Identity management and user
authentication are a concern for some Government officials because
customers and a provider may need to establish a means to securely
exchange and rely on authentication and authorization information for
system users. In addition, responding to security incidents may be more
difficult in a shared environment because there could be confusion over
who performs the specific tasks--the customer or the provider.
Although there are numerous potential information security risks
related to cloud computing, these risks may vary based on the
particular deployment model. For example, NIST stated that private
clouds may have a lower threat exposure than community clouds, which
may have a lower threat exposure than public clouds. Several industry
representatives stated that an agency would need to examine the
specific security controls of the provider the agency was evaluating
when considering the use of cloud computing.
federal agencies and government-wide initiatives had begun to address
information security issues for cloud computing, but remained
incomplete
In our report, we noted that Federal agencies had begun to address
information security for cloud computing; however, they had not
developed corresponding guidance. About half of the 24 major agencies
reported using some form of public or private cloud computing for
obtaining infrastructure, platform, or software services. These
agencies identified measures they were taking or planned to take when
using cloud computing. These actions, however, had not always been
accompanied by development of related policies or procedures.
Most agencies had concerns about ensuring vendor compliance and
implementation of Government information security requirements. In
addition, agencies expressed concerns about limitations on their
ability to conduct independent audits and assessments of security
controls of cloud computing service providers. Several industry
representatives were in agreement that compliance and oversight issues
were a concern and raised the idea of having a single Government entity
or other independent entity conduct security oversight and audits of
cloud computing service providers on behalf of Federal agencies.
Agencies also stated that having a cloud service provider that had been
precertified as being in compliance with Government information
security requirements through some type of Government-wide approval
process would make it easier for them to consider adopting cloud
computing. Other agency concerns related to the division of information
security responsibilities between customer and provider. As a result,
we reported that the adoption of cloud computing by Federal agencies
may be limited until these concerns were addressed.
Several Government-wide Cloud Computing Information Security
Initiatives Had Been Started, but Key Guidance and Efforts Had
Not Been Completed
In our May 2010 report, we also noted that several Government-wide
cloud computing security activities had been undertaken by
organizations such as the Office of Management and Budget (OMB),
General Services Administration (GSA), the Federal Chief Information
Officers (CIO) Council, and NIST; however, significant work remained to
be completed. Specifically, OMB had stated that it had begun a Federal
cloud computing initiative in February 2009; however, it did not have
an overarching strategy or an implementation plan. In addition, OMB had
not yet defined how information security issues, such as a shared
assessment and authorization process, would be addressed.
GSA had established the Cloud Computing Program Management Office,
which manages several cloud computing activities within GSA and
provides administrative support for cloud computing efforts by the CIO
Council. The program office manages a storefront, www.apps.gov,
established by GSA to provide a central location where Federal
customers can purchase software as a service cloud computing
applications. GSA had also initiated a procurement to expand the
storefront by adding infrastructure as a service cloud computing
offerings such as storage, virtual machines, and web hosting. However,
GSA officials reported challenges in addressing information security
issues as part of the procurement. As a result, in early March 2010,
GSA canceled the request and announced plans to begin a new request
process. GSA officials stated that they needed to work with vendors
after a new procurement was completed to develop a shared assessment
and authorization process for customers of cloud services purchased as
part of the procurement, but had not yet developed specific plans to do
so.
In addition to GSA's efforts, the CIO Council had established a
cloud computing Executive Steering Committee to promote the use of
cloud computing in the Federal Government, with technical and
administrative support provided by GSA's Cloud Computing Program
Management Office, but had not finalized key processes or guidance. A
subgroup of this committee had developed the Federal Risk and
Authorization Management Program (FedRAMP), a Government-wide program
to provide joint authorizations and continuous security monitoring
services for all Federal agencies, with an initial focus on cloud
computing. The subgroup had worked with its members to define
interagency security requirements for cloud systems and services and
related information security controls. However, a deadline for
completing development and implementation of a shared assessment and
authorization process had not been established.
NIST is responsible for establishing information security guidance
for Federal agencies to support the Federal Information Security
Management Act of 2002 (FISMA); however, at the time of our report, it
had not yet established guidance specific to cloud computing or to
information security issues specific to cloud computing, such as
portability, interoperability, and virtualization. The NIST official
leading the institute's cloud computing activities stated that existing
NIST guidance in Special Publication (SP) 800-53 and other publications
applied to cloud computing and could be tailored to the information
security issues specific to cloud computing. However, both Federal and
private sector officials had made clear that existing guidance was not
sufficient.
agencies have made progress in implementing gao recommendations, but
additional actions are needed to assist agencies in securely
implementing cloud computing
In our May 2010 report, we made several recommendations to OMB,
GSA, and NIST to assist Federal agencies in identifying uses for cloud
computing and information security measures to use in implementing
cloud computing. These agencies generally agreed with our
recommendations. Specifically, we recommended that the Director of OMB
establish milestones for completing a strategy for implementing the
Federal cloud computing initiative; ensure the strategy addressed the
information security challenges associated with cloud computing, such
as needed agency-specific guidance, the appropriate use of attestation
standards for control assessments of cloud computing service providers,
division of information security responsibilities between customer and
provider, the shared assessment and authorization process, and the
possibility for precertification of cloud computing service providers;
and direct the CIO Council Cloud Computing Executive Steering Committee
to develop a plan, including milestones, for completing a Government-
wide security assessment and authorization process for cloud services.
In response, in February 2011, OMB issued its Federal Cloud
Computing Strategy,\7\ which references the establishment of a shared
assessment and authorization process for cloud computing. In addition,
the strategy discusses other steps to promote cloud computing in the
Federal Government, including ensuring security when using cloud
computing, streamlining procurement processes, establishing standards,
recognizing the international dimensions of cloud computing, and
establishing a governance structure. However, the strategy does not
address other security challenges such as needed agency-specific
guidance, the appropriate use of attestation standards for control
assessments of cloud computing service providers, and the division of
information security-related responsibilities between customer and
provider. Until these challenges are addressed, agencies may have
difficulty readily adopting cloud computing technologies.
---------------------------------------------------------------------------
\7\ OMB, Federal Cloud Computing Strategy (Washington, DC: February
2011).
---------------------------------------------------------------------------
We also recommended that the Administrator of GSA, as part of the
procurement for infrastructure as a service cloud computing
technologies, ensure that full consideration be given to the
information security challenges of cloud computing, including a need
for a shared assessment and authorization process.
In response, GSA issued a request for quote relating to its
procurement for cloud services that included the need to use FedRAMP
once it is operational. FedRAMP was further developed by GSA, in
collaboration with the Cloud Computing Executive Committee, as a shared
assessment and authorization process to provide security authorizations
and continuous monitoring for systems shared among Federal agencies.
The CIO Council, in collaboration with GSA, issued a draft version of
the shared assessment and authorization process in November 2010;\8\
however, the process has not yet been finalized. GSA officials stated
that they intend to release additional information on FedRAMP once OMB
issues a policy memorandum related to cloud computing, expected in the
first quarter of fiscal year 2012.
---------------------------------------------------------------------------
\8\ CIO Council, Proposed Security Assessment and Authorization for
U.S. Government Cloud Computing, Draft version 0.96 (Washington, DC:
November 2010).
---------------------------------------------------------------------------
Last, to assist Federal agencies in implementing appropriate
information security controls when using cloud computing, we
recommended that the Secretary of Commerce direct the Administrator of
NIST to issue cloud computing information security guidance to Federal
agencies to more fully address key cloud computing domain areas that
are lacking in SP 800-53, such as virtualization, data center
operations, and portability and interoperability, and include a process
for defining roles and responsibilities of cloud computing service
providers and customers.
NIST has also taken steps to address our recommendations. In
January 2011, it issued SP 800-125, Guide to Security for Full
Virtualization Technologies.\9\ Virtualization is a key technological
component of cloud computing. SP 800-125 discusses the security
characteristics of virtualization technologies, provides security
recommendations for virtualization components, and highlights security
considerations throughout the system life cycle of virtualization
solutions. In July 2011, NIST issued SP 500-291, NIST Cloud Computing
Standards Roadmap,\10\ and in September 2011, SP 500-292, NIST Cloud
Computing Reference Architecture.\11\ Collectively these documents
provide guidance to help agencies understand cloud computing standards
and categories of cloud services that can be used Government-wide.
Among other things, these publications address cloud computing
standards for interoperability and portability.
---------------------------------------------------------------------------
\9\ NIST, Guide to Security for Full Virtualization Technologies,
SP 800-125 (Gaithersburg, MD: January 2011).
\10\ NIST, NIST Cloud Computing Standards Roadmap, SP 500-291
(Gaithersburg, MD: July 2011).
\11\ NIST, NIST Cloud Computing Reference Architecture, SP 500-292
(Gaithersburg, MD: September 2011).
---------------------------------------------------------------------------
NIST also issued a draft publication on cloud computing, SP 800-
144, Guidelines on Security and Privacy in Public Cloud Computing,\12\
which addresses the security concerns associated with data center
operations and the division of responsibilities among providers and
customers. In addition, the guide discusses the benefits and drawbacks
of public cloud computing, precautions that can be taken to mitigate
risks, and provides guidance on addressing security and privacy issues
when outsourcing support for data and applications to a cloud provider.
According to NIST officials, SP 800-144 will be finalized in the first
quarter of fiscal year 2012.
---------------------------------------------------------------------------
\12\ NIST, Guidelines on Security and Privacy in Public Cloud
Computing, Draft SP 800-144 (Gaithersburg, MD: January 2011).
---------------------------------------------------------------------------
In summary, the adoption of cloud computing has the potential to
provide benefits to Federal agencies; however, it can also create
numerous information security risks. Since our report, Federal agencies
have taken several steps to address our recommendations on cloud
computing security, but more remains to be done. For example, OMB has
issued a cloud computing strategy; however the strategy does not fully
address key information security challenges for agencies to adopt cloud
computing. The CIO Council and GSA have also developed a shared
assessment and authorization process, but this process has not yet been
finalized. In addition, NIST has issued several publications addressing
cloud computing security guidance. Although much has been done since
our report, continued efforts will be needed to ensure that cloud
computing is implemented securely in the Federal Government.
Chairman Lungren, Ranking Member Clarke, and Members of the
subcommittee, this concludes my prepared statement. I am pleased to
respond to any questions.
Mr. Lungren. Thank you very much.
Thank all three of you for that. I understand we are going
to have votes in about 10 minutes so we will see if we can get
through a couple of 5-minute question periods. I will start.
If I were to summarize what I heard, it is that Mr. Spires
and Mr. McClure have the glass-half-full approach, and Mr.
Wilshusen, you have the glass-half-empty approach.
Mr. Spires and Mr. McClure, can you tell me which glass I
should take up?
Mr. Spires. Well, sir, I do have the glass-half-full
approach. I believe that cloud computing is going to transform
I.T. as things become more commoditized. The world is moving
that way; we need to move with it because the advantages are so
great.
Mr. Lungren. So it is inevitable that we are going to move
there?
Mr. Spires. I think it is inevitable.
Mr. Lungren. So the question we have here is: How secure
can we make it?
Dr. McClure, you--if I were to just listen to what you had
to say I would be very, very pleased that it is very secure
right now or on the process of getting even more secure. But
the gentleman to your left is paid to poke holes in arguments
that people like you make, and he has poked some holes.
Sometimes things sound too good to be true, and most of the
time I have found that is true. What assurance do we have as we
move toward this cloud computing--well, let me put it this way:
In the report that we issued yesterday, and this is consistent
with what we have heard before this committee, there has been
the suggestion that 85 percent of computer intrusions,
unwarranted interference, et cetera, could be stopped by good
computer hygiene, which suggests that we have a lot to do in
terms of public and private awareness.
One of the key aspects to security on cloud computing would
be awareness. How am I to be able to tell my colleagues and my
constituents that the awareness that evidently isn't there now
with the way we are doing things is going to be there as we
move to computing? Because isn't that the essential question?
You can set up the best sort of secure systems possible,
but if there is not the awareness of what you have to do, both
in terms of what we are talking about here, the ultimate user,
that is, the Government employee, but also the vendor, and the
vendor's employees--it is not going to happen. So is that
computed into what you said today, that we have the awareness,
we are going to have the awareness, it is built in, or it is
easier in a cloud computing atmosphere than what we have had
thus far?
Mr. McClure. Well, thank you, Mr. Chairman. As Greg knows,
I used to be a hole-poker, as well, because I sat in GAO, so
there is no--this is a really challenging area, so I don't
think it is a half-full, half-empty glass. We are never done in
this area. I think all of us here at the table would agree with
that.
We can put the best controls in place, the best policies,
the best people, but you are going to always be advancing in
your knowledge and in your ability to deter threats and
vulnerabilities to your system. So it is a given.
So I think that is one thing we need to do is to dispel the
myth that there is some magical control or formula that we are
not using and if we just put in place we would--we will be
absolutely secure. Security is an on-going exercise.
Mr. Lungren. True. But how do we answer the question to
those who would be skeptics of what we think we need to do,
that if you move in the direction of cloud computing you are
necessarily creating greater target-rich environments? That is,
if I can invade a cloud that has multiple--more data points
than a small network I would target my energies on that, and if
I am successful, boy, I really have a tremendous amount of
information, and connected information, where I may not have it
if it is divided over 2,700 different networks. That is the
concern I have expressed to me.
On the other hand, I hear the argument, ``Well, wait a
second. We can put more capital investment into cloud
technology. They can be more up-to-date, more timely. They can
find things more quickly because they have a greater
observation point.'' I understand that.
But I think you understand the point about a greater
target-rich environment with the concern people then have that
you have got to have a promise that the security of the cloud
is going to be measurably better than the security we have in
the current system.
Mr. McClure. Yes, and I would absolutely agree, that is the
way forward. Our problems in security are not unique cloud
computing systems, by the way. So if you look at what we are
putting in place in FedRAMP, we need, first of all, agreement
on what the baseline controls actually are, and I think we have
achieved that by working across a huge community in the
Government to have that dialogue.
Second, we have to agree on what are the additional
controls that are warranted in a cloud environment, much as you
described, where there are extended vulnerabilities that are
not necessarily applicable to traditional systems. So we have
done that. We have tried to introduce new controls.
Third, we have to move to continuous monitoring. We have to
make sure that agencies are applying managerial, technical, and
operational controls to their systems for clouds, but we also
have to report on a real-time basis the posture of the cloud
security provider's environment, and that we have to see and we
have to be able to take action, and we have to demand a
solution be put in place. Then we can really bump up, I think,
our security posture to more tolerable levels.
Mr. Lungren. Thank you very much.
Now, I either recognize the gentlelady or Mr. Thompson.
No, whoever you want to----
Mr. Thompson. Well, thank you very much.
Mr. Lungren. Because we have, I think, 5 minutes, probably,
before we have to go vote. Votes have already been called. So--
--
Mr. Thompson. Right. Well, thank you, Mr. Chairman. I
appreciate the Ranking Member's indulgence.
Clearly, the cloud is kind of cloudy right now to a lot of
us, and we are trying to get better. But as we go forward, I am
a little concerned about how our Government moves forward
without the necessary safeguards in place.
Mr. Spires, let us talk about one of my concerns. I
understand that DHS has contracted with a company called CGI
Federal, Inc., to move its public website to the cloud. Now, I
understand that this is not a U.S. company. Am I correct or
incorrect?
Mr. Spires. Actually, sir, CGI Federal--well, you are
correct, we are--we have contracted, through the GSA
infrastructure as a service vehicle for CGI Federal to provide
cloud services so we can move our public-facing websites to the
cloud. That is correct.
CGI Federal is a U.S.-based company. The parent company is
a Canadian-based company.
Mr. Thompson. So it is a U.S.-based company----
Mr. Spires. Yes, sir.
Mr. Thompson [continuing]. Owned by a Canadian company?
Mr. Spires. That is correct, sir.
Mr. Thompson. Okay. Does that cause you any concern?
Mr. Spires. In awarding the contract, sir, and going
through the evaluation, we followed all the proper regulations
from the FARR. I worked with our procurement organization,
worked with GSA's procurement organizations.
I should also point out, sir, that we put a clause into
that contract or that task order that States that everyone that
works on that particular contract needs to be a U.S. citizen
unless we grant a waiver, and I don't expect we would be
granting a waiver to that, and that all the data that is--that
we would use in running those public websites needs to be
resident within the United States.
Mr. Thompson. Can you provide the committee with a copy of
that task order?
Mr. Spires. We certainly can, sir.
Mr. Thompson. So none of the work--none of the hosting or
anything will be done out of the----
Mr. Spires. No. The hosting will be done in two geographic
diverse data centers that are both located within the United
States, sir.
Mr. Thompson. Thank you very much.
Dr. McClure, when you testified before the House Oversight
and Government Reform Committee last year you called security
one of the most significant obstacles to the adoption of cloud
computing. Is that still your position or have you modified it?
Mr. McClure. No, and I think it is a--the top challenge.
There are others that we have alluded to.
Security, because of these issues we have been bringing up
this morning--the lack of consistent standards, the lack of the
quality of the work being done to assess cloud systems, the
lack of real continuous monitoring, real-time capabilities--it
presents real challenges, particularly in cloud environments.
But we are addressing those; that is what we are trying to do.
The other two, though--and I think Greg may have mentioned
this--are portability--I park my data onto a cloud provider's
system; I, either by choice or because they are going out of
business, I want to get that data off of their cloud system and
into a new one. Can they aggregate and reconstitute that data
and give it back to me? It is a huge question that Federal
officials have to ask of their cloud service provider.
Mr. Thompson. So that is still a concern?
Mr. McClure. Absolutely.
Mr. Thompson. Well, I understand that we have 12 companies
that have been approved for some services under these
contracts, while only four have been--of those 12--have been
fully vetted. Is there some issues around security, or what?
Mr. McClure. Absolutely. Once the 12 entities were found to
be qualified and awarded business under that BPA, the second
step is to go through a security authorization process, which
is controls and testing to make sure they meet all Federal
requirements. To date, four have, and they are subcontractors,
and the remaining are going through the completion of that
security authorization.
Mr. Thompson. So another Federal agency couldn't pick from
the eight at this point?
Mr. McClure. Correct.
Mr. Thompson. They can only take the four?
Mr. McClure. They can take the four. They can actually, if
they wanted to, enter into business with one of the other eight
if they themselves performed the security assessments. We are
doing it at GSA in order for all agencies to be leveraging off
of that rather than repeating it.
Mr. Thompson. Well, and I guess for the GAO person in my
last second, I am a little concerned that some of the vetting
is not complete with some of the companies. Have you looked at
that and whether or not you have some concerns around that,
also?
Mr. Wilshusen. Well, we haven't specifically looked at
GSA's authorization and assessment process yet, but certainly
if we haven't--or the GSA or Federal agencies have not yet
assessed the security controls over the cloud environment, they
are doing--if they use that environment they are doing so at
risk, and at an increased risk.
Mr. Thompson. Yes. Thank you.
Mr. Lungren. All right. We are expected at a series of five
votes on the House floor that has already started. We have, I
think, 5 minutes to get over there to vote.
The subcommittee will stand in recess until the conclusion
of these votes, reconvene immediately following the last vote,
which will probably be between 45 minutes to an hour.
[Recess.]
Mr. Lungren. With the acceptance of the Minority I am going
to ask a few questions, and then, when Ms. Clarke gets here she
will have the chance, or Mr. Keating returns from the floor, so
we can allow the first panel to go as quickly as possible.
Let me ask you, Mr. Spires, how is the Department
evaluating the different needs for different data sets? That
is, if we have an agreement that there are different categories
of clouds that are appropriate for different levels of security
based on the nature of the data, what is the criteria you are
using in evaluating those different needs?
Mr. Spires. My apologies. Yes, sir. We are using different
evaluation--or, using evaluation criteria based on the
sensitivity of the data itself. So in our case, we are starting
off fairly simple right now.
All of what we would consider sensitive data, including
data that would be for official use only and higher sensitivity
data--law enforcement sensitive, for instance, in the
unclassified realm--right now we are keeping that within what
we call our private cloud, and that private cloud is hosted out
of our two enterprise data centers. It runs within our own
wide-area network, and hence, we are able to control that
environment and really have the insights through continuous
monitoring into the security stature of that environment.
We are aggressively looking at public cloud for what we
would say is nonsensitive data. So the example I used in my
testimony of us moving our public-facing websites, like
dhs.gov, fema.gov, to public cloud, and we are trying to get
experience using the public cloud.
As the FedRAMP process matures we would anticipate over
time looking at how that evaluation criteria could change,
because I am a real believer, having been in the private sector
for a good part of my career, that we always want to foster
competition; we always want to have choice. So as we have more
and more comfort over time that public cloud services can
provide the security levels, okay, and the continuous
monitoring capabilities that we need we would look, then, over
time to start to relax that criteria or shift it so that more
sensitive data would be able to be moved into the public cloud.
Mr. Lungren. Now, what is the interplay between Department
of Homeland Security and GSA in terms of assurance of
cybersecurity as we move to the cloud? DHS appears to be the
point agency for--I don't want to say looking over the
shoulder, but looking at other Government agencies and
departments to assure that they are taking cybersecurity
seriously. I know we have the office in the White House, which
is an office that I would suggest is sort of a--my definition,
sort of a focal point for policy, but DHS is the operational
point.
How do you interface with GSA on something like this, with
respect to their responsibilities in the areas that they have
authority?
Mr. Spires. Let me provide an answer, and I am sure Dr.
McClure will then want to weigh in.
First, I should state that I am the CIO for the Department
of Homeland Security; there is another part of DHS within what
we call our NPPD organization that really has this mission, if
you will, to provide--really look at cybersecurity, of course,
for the Nation, but in particular, for the civilian government
agencies.
Mr. Lungren. Hopefully you folks talk to one another.
Mr. Spires. We talk to one another all the time. As I like
to say, we are the biggest guinea pig for what they want to do
next. I think we should be, right? So we work very, very
closely with them.
So they have, for instance the US-CERT operation, which
gathers----
Mr. Lungren. Right.
Mr. Spires [continuing]. Incident response information from
throughout the Government to be able to share, analyze that
information. That organization is working very closely with our
organization and with GSA as we look at how we are going to
roll out this FedRAMP initiative.
For instance, as FedRAMP rolls out and we look at
continuous monitoring for public cloud service providers, those
feeds would be provided to the Department of Homeland Security,
to US-CERT, for continual analysis, as well as to the agency,
so that we can continue to monitor, if you will, public cloud
capabilities, if you will, real-time throughout the Government
for the use of the public cloud.
Mr. Lungren. Dr. McClure.
Mr. McClure. Yes, it is a--excuse me, Mr. Chairman--it is a
very complementary relationship. FedRAMP has actually been
devised with heavy DHS participation, both from Richard's
office, representing the CIO angle, and from Greg Schaffer's
office, the NPPD directorate that Richard referred to, which
does the operational monitoring and runs a lot of the--a lot of
the US-CERT capabilities.
So what we are doing in FedRAMP is designed to actually
incorporate the role of DHS into that process. We are not
replicating, we are not eliminating anything that is really
clearly in DHS space.
In fact, if you look at the recent change made to FISMA
that requires agencies to do monthly reporting of continuous
monitoring, FedRAMP is simply building on top of that. It is
utilizing that process as we designed our process for FedRAMP.
Mr. Lungren. In either your opening statement or an answer
to a question you indicated that continuous monitoring was one
of the essentials as we move to cloud computing. Is the
suggestion that this needs to be increased in intensity? Is it
a relatively new concept? Is it one that has been implemented
across the board in Government agencies and departments, or is
it sporadic?
Given what you said about this being an essential, one
would think it would be essential now, and one would also ask
whether it is treated as something essential now.
Mr. McClure. Absolutely. The issue with the continuous
monitoring controls is the agreement upon the standard for the
control and on the data elements that actually would be passed
to show compliance.
What we want to do is to make sure that that has been
agreed to with industry as well as inside of Government. So
that is the process that is underway now, establishing those
standards for the controls and the continuous monitoring are
and coming up with agreement on the actual data elements that
would be shared between entities to show compliance.
Once that is worked out, I think we can begin moving to a
near real-time view of what is happening in the provider space,
whether it is an internal or external provider that is doing--
--
Mr. Lungren. Mr. Wilshusen, do you have any comments,
please?
Mr. Wilshusen. Yes, I do. Thank you very much.
As you know, we issued a report just this week on Federal
information security. One of the issues we discuss had to do
with continuous monitoring.
It is a relatively new phenomenon and requirement within
the Federal Government. NIST recently issued some guidance that
included it in its risk framework. I believe that came out back
in February, perhaps, of 2011, or--I think it was February
2011, if I remember correct.
Right now the experience with Federal agencies in
continuous monitoring is still immature, if you will. There is
still a great deal that needs to be done. In some respects it
is required that agencies have the capability to have automated
tools in place in which they can gather this information and
feed it on a regular near-real-time basis, and many of the
agencies so far don't have those capabilities over all of their
assets.
It is also important to know that with continuous
monitoring there is that automated aspect of it, but there is
still a need for testing and evaluation of the effectiveness of
the controls to assure that the information that is being
provided through these automated tools is accurate and
reliable.
Mr. Lungren. One of the key risks the GAO report identifies
relating to cloud computing is the dependency on vendor. There
was mention by Mr.--by Dr. McClure when we were doing the first
round of questions about the scenario in which you terminate a
contract or a vendor ceases operations.
Any thoughts on how you protect against the vulnerability
there? What do you have to build in to protect the Government's
essential needs at that point?
Mr. Wilshusen. Well, that certainly is a key risk to
Federal agencies. When we did our report last May all 24 of the
24 agencies cited loss of information as a key risk should
their cloud service be terminated.
So in terms of being able to help mitigate those risks, it
is imperative for agencies to establish comprehensive service-
level agreements that specify clearly up front what the roles
and responsibilities of the cloud service provider is as well
as what the customer is with regard to providing information
should they go out of business. It is also--or service is
discontinued.
It is also imperative that interoperability and portability
standards be developed and implemented so that agencies have
the capability to take their information that is being
processed by a cloud service provider and use it either
internally or to another provider should the need arise.
Mr. Lungren. Mr. Spires, is there anything technologically
unique about cloud computers that causes more difficulty with
this particular concern--that is, termination of services?
Mr. Spires. Not on the technical side, sir, but I would
echo what Greg said, that one of our big concerns about moving
to the public cloud is exactly that, that we want to be able to
assure continuity of service to our customers, right, in all
events. So we have to work those scenarios as to what happens
in the hopefully unlikely event that that cloud service
provider can no longer offer that service--so data archiving
capabilities, having the standards set--and I know this is
something NIST is working on--for cloud interoperability so
that we can quickly shift to another cloud service provider if
necessary.
Mr. Lungren. So cloud interoperability would presume that
you have equal security measures available.
Mr. Spires. Well, I think that comes back to the FedRAMP
initiative and the idea of having these provisional
authorizations in place for, hopefully over time, many cloud
service providers so that that makes it much easier for us, as
CIOs, to have choice and to be able to much more easily move
our services. Goes back to my competition point earlier. It
also gives us a more competitive playing field, which will
drive down costs over time and, of course, provide better
service.
Mr. Lungren. Before I yield to the gentlelady, the Ranking
Member, the Ranking Member of the full committee brought up the
question about the contract with the first that is a U.S.-based
firm but a wholly-owned subsidiary of a Canadian firm. We are
close to Canada, but it is another country, as I recall.
I think Congressman Thompson was bringing up the question
of the--I don't know the visuals of that or how we tell the
American people, ``Yes, we are going to have--the Government is
going to use vendors that have cloud computing with all of the
assets but also the vulnerabilities we talked about, and it is
going to be a company that answers to people who aren't in this
country.''
You answered it specifically. Do you understand the--at
least the question some people might have there?
Mr. Spires. Sure. The more general point--certainly at the
Department of Homeland Security within my office, we would be--
want to always make sure, sir, that our data is protected, that
for any sensitive data as we move forward that we would want
U.S. citizens to only have access to that data, that it be
housed--for sensitive information, that we would only have that
data housed in data centers that were on American soil. That
would go without--I mean, it is the given, okay?
All I can say, sir, is we followed the regulations. We did
an open competition within the providers that were available to
us through the GSA vehicle, and based on the evaluation
criteria, this firm won that particular task order.
Mr. Lungren. Okay. Thank you very much.
The gentlelady, the Ranking Member of the subcommittee, is
recognized for 5 minutes.
Ms. Clarke. Thank you very much, Mr. Chairman.
Let me thank our panelists and thank you for your patience.
We need clones around here, that is all I can say.
But let me say that in the brief moments I have had in the
hearing I am not as concerned about our capability to secure
the cloud, and I say that simply because we were innovative
enough to invent it. I believe that our knowledge, our
capability, our skills will enable us to protect. So I am going
to be affirmative.
Then when I think about young people today and their level
of curiosity, their innovativeness, I know that somewhere
seated in some classroom today is the person that is going to
come forth who will enable us to do what we need to do to move
forward with the innovations that we have as a civil society.
So I am coming at this not as a scary person but as someone who
is ready for the adventure.
Having said that, I would like to ask this question of both
Mr. Spires and Mr. Wilshusen: Did you look at the experiences
of other Federal agencies in using public clouds before
undertaking this effort? If so, what lessons did you learn and
how did you apply them? What about State and private sector
experiences? Were those also taken into account?
Mr. Spires. Ma'am, we certainly have, within our strategy,
had numerous discussions, both with other Federal Government
agencies--NASA, the Veterans Administration come to mind, both
of which have been very aggressive at looking at cloud
capabilities. We have also talked to a number of--I have
personally talked to a number of CIOs within private sector
firms as well as my staff, who have been very involved in
reaching out, as well as to advisory services that work in the
I.T. industry and serve that industry.
A few of the lessons learned--and I think we are still
learning a lot of these lessons, right? I mean, one of our
biggest issues, beyond security, because that is probably the
biggest issue; we have been talking about that. But the next
one is really, this is fundamentally a different business
model, and it changes--I mean, we are buying a service-level
agreement; we are not, you know, out there purchasing hardware
and licensing software and integrating together.
Fundamentally, how we procure this is very, very different.
So we have been working across the Federal Government--and as a
matter of fact, in a couple weeks the Federal CIO Council and
the Federal Chief Acquisition Officer Council are going to be
meeting together to talk about this very issue: How do we work
out the procurement issues, the business model issues, so that
we put ourselves in the best position to leverage this
capability from a business perspective?
I would say that is where a lot of the lessons learned are.
I think many of us are still feeling our way, to be honest, as
to what is the right business model moving forward.
Mr. Wilshusen. Ms. Clarke, when we conducted our review
last year over cloud computing security we went to a couple of
different agencies and looked at some of the pilot cases that
were underway. We went to DOD and looked at the DISA RACE,
which is the Rapid Access Computing Environment, and also
looked at NASA's Nebula cloud environment.
A couple of lessons learned that they experienced had to do
with just the assuring that they are having to reengineer some
of their business processes in order to accommodate the use of
the cloud computing. They also found that one of the challenges
that they had was also clearly specifying and delineating the
responsibilities for security of the client personnel, you
know, at NASA, as well as the cloud provider.
Now, in both cases each of their implementations were
private cloud implementations. They decided in each case to
take a kind of a slow, cautious approach before jumping in and
maybe going to a public cloud. But in both cases they went to
the private cloud implementation, which generally will have a
lower threat exposure than public cloud.
Ms. Clarke. Then I want to ask, are there any agency
applications or services that should never move to the cloud,
or is everything an agency does open to the move? In either
case, why would it be the case?
Mr. Wilshusen. Well, I will take an initial stab at that.
There is probably implementations and information that is so
sensitive, perhaps, you know, classified information that needs
to be particularly protected that it should not be placed out
into a cloud environment, particularly a public cloud
environment, given the current security capabilities present.
So certainly classified information probably should not be
placed in a public cloud environment.
Ms. Clarke. So would you say never, or do you foresee in
the future that that capability will exist? Because my question
was never.
Mr. Wilshusen. Right. Well, I was taught from a very early
age never to say never, and I think I will keep to that now.
Mr. Spires. I think I have essentially the same answer, Ms.
Clarke. In the I.T. field I have learned to never say never
because things change so much, since, certainly, in the years I
have been in this field.
That being said, I would agree wholehearted with Greg. It
is going to be quite a while before we would have any comfort
in putting any classified information into a public cloud
environment, and it may never happen. I think it will quite a
few years before we would look to do that.
Mr. McClure. Yes. The only thing I would add, Ms. Clarke,
is that it goes back to what the agency sets as its
requirements for what it is trying to do with its data and its
service delivery. If the data demands protection levels that
are beyond the capabilities of either in-house or out-house
providers then you have got to address that.
So the term ``public cloud'' is used pretty loosely.
Actually, there are instances, I think, where you will see
Federal agencies claiming they do have things in public cloud
but it is not the equivalent of what you might find a consumer
such as ourselves doing from our own homes.
We have security requirements, records management
requirements, 508 requirements. They have all these other
requirements that still these providers have to show that they
are able to provide that even though they may be called a
public cloud solution.
Ms. Clarke. Thank you very much, Mr. Chairman.
Mr. Lungren. Mr. Keating, you are recognized for 5 minutes.
Mr. Keating. Thank you, Mr. Chairman.
With the new technologies I think there is a possibility of
increased risk on infringement of copyright holders' rights
because of the nature of this, that it is faster, cheaper, and
it is easier to engage with unauthorized reproduction and
distribution of public performances of types of copyrighted
works. To what extent can the increased reliance on the data
storage through cloud computing services contribute to this
kind of copyright infringement? Do you see an issue there?
I will throw it open to the whole panel.
Mr. McClure. Sure. I will take a stab at it first.
I think it goes back to in any environment, private or
public cloud regardless, you have still basic security and
privacy standards that have to be met. Access controls come to
mind in this particular case. Who has access to information in
these cloud environments is still a huge issue. If you don't
define that and put the controls in place then you are subject
to losing information no matter what kind of cloud environment
you have it in.
Mr. Keating. Yes.
Anyone else?
Mr. Spires.
Mr. Spires. I would just add, sir, that one of the things
we are really working on within Homeland Security is
strengthening our identity credential and access management
capabilities, to pick up on what Dr. McClure said. We foresee
in the future having a much stronger authentication model to
protect against these very types of things, whether it be
copyright infringement, or in our case we are very concerned
about privacy and civil liberties, right, and access to the
data that we store.
That really transcends whether you are in a cloud
environment or whether this is just a more traditional kind of
I.T. system and database. But these are the things that we are
working on right now that strengthen the safeguarding side yet
still enable the right kind of information-sharing to protect
the homeland.
Mr. Keating. Okay.
Mr. Wilshusen. I would just like to add that I agree that
authorization and identification and verification is going to
be key in this respect. The one additional wrinkle--not to poke
a hole or anything--is that the responsibility for sharing that
the authorization is correct and the identity of the user is
actually verified and claimed may no longer reside with the
Federal Government or the Government agency with the cloud
service provider. So the effectiveness of the cloud service
provider's controls and access controls come into play as well.
Mr. Keating. Okay. That is interesting.
Thank you very much. I had just one other--might be a bit
tangential, but, you know, in terms of the Government security
and securing Government data, there is the use of flash drive-
type products as well. Is there any advantage or
differentiation that is being made when you have that kind of,
you know, product, in using a hard drive kind of system versus
a software authentication?
Do you get anything more out of--from a secure basis--out
of the hardware kind of authentication for that type of product
than just the software itself? I mean, is it--where do you see
it going? I mean, do you need both? Is it fine just with
software, or do you think there is a need for that going
forward for secure data?
Mr. Wilshusen. Well, I will take the initial stab. Yes, I
think, you know, the hardware's authentication and security is
something that can definitely help protect information, and
particularly with flash drives and thumb drives. It, as you
know, is a key risk because those devices can contain----
Mr. Keating. Right.
Mr. Wilshusen [continuing]. Large volumes of information
and they are extremely portable, as they are designed to do.
Some agencies, like the Department of Defense, has banned their
use on their systems because they also are carriers or can be
used to carry malicious software and install that on devices on
an agency's internal network.
Mr. Keating. Okay.
I will yield back my time, Mr. Chairman. Thank you.
Mr. Lungren. Thank you very much.
I want to thank this first panel for not only testifying
but understanding we have votes that interrupt, I understand
that this takes a portion of your day, and we appreciate you
being here. We thank you for your testimony.
The Members would request the Members of the committee may
have some additional questions for you that we might submit in
writing. We would ask that you would respond to those in
writing.
With that, I am happy to thank you and dismiss you, and we
will move on to the second panel.
I am going to ask unanimous consent that Mr. Duncan, who is
a Member of the full committee but not a Member of the
subcommittee, can sit for this second panel and have the
privilege of introducing someone from his State when we get
there.
So thank you, to the first panel.
If the second panel would come up, Mr. Sheaffer, Mr. Brown,
Mr. Bottum, and Mr. Curran?
Today we have the opportunity to hear from a distinguished
second panel on the question of ``Cloud Computing: What Are the
Security Implications?''
We have Mr. James Sheaffer, the president of Computer
Science Corporation's North American Public Sector. Previously,
Mr. Sheaffer served as vice president for CSC as well as a
general manager of Prime Alliance--that is CSC's partnership
with the IRS--to support the business systems modernization
program. Prior to joining CSC Mr. Sheaffer spent 27 years in
the American Management Systems, Inc. working on
telecommunication in North America and Europe.
Mr. Timothy Brown is vice president and the chief architect
for security management at CA, Inc. With more than 20 years of
information security experience, Mr. Brown has been involved in
many areas of security, including threat research,
vulnerability management, consumer and enterprise identity,
access management, network security in the encryption
compliance and managed security services.
John Curran is the president and CEO of American Registry
for Internet Numbers. He serves as the chief technology officer
and chief operating officer for ServerVault as well as the
chief technology officer at XO Communications and BBN/GTE
Internetworking. Mr. Curran also has been an active participant
in the Internet Engineering Task Force.
It is my privilege to allow Mr. Duncan to introduce the
next gentleman, who, as I understand, had something to do with
Purdue University. Since I went to Notre Dame I would like you
to introduce him.
Mr. Duncan. Okay. Thank you.
[Laughter.]
Mr. Duncan. Thank you, Mr. Chairman. Thanks for giving me
the opportunity, and thanks to the committee for allowing me to
sit on the dais with you this morning.
It is my distinct pleasure to introduce one of my
constituents, but he is also someone from my alma mater,
Clemson University. Jim Bottum is a chief information officer
and vice provost for computing and information technology for
Clemson University.
Clemson, Mr. Bottum leads efforts focusing on high-
performance computing and communication as well as
collaborating with State and National government entities.
Under his leadership, Clemson University's Palmetto Cluster has
appeared at No. 60 in the world's top 500 computing sites
alongside Clemson's Computational Center for Mobility Systems,
ranked at No. 100.
Mr. Bottum currently serves on the NSF Advisory Committee
for Cyber-Infrastructure, NSF Advisory Committee for CRPA
Assessment, and the I-2 or Internet 2 Board of Trustees. Prior
to coming to Clemson, Mr. Bottum was the first CIP and VP for
computing at Purdue, where he was responsible for planning and
coordinating all computing and information systems across the
university.
He has also served on other NSF committees as well as
National laboratory boards and provided consulting services for
major universities across the United States. He has worked
extensively on issues of cloud computing and should provide an
excellent perspective of this issue from his academic research
and experience.
I look forward to hearing his testimony and thank you for
having him here today. I yield back.
Mr. Lungren. I thank the gentleman.
We thank you all for being here. We thank you for your
indulgence, in that I know you had to wait as well, as we went
over to vote.
We have the procedure here that your written remarks will
be made a part of the record in their entirety, and we would
ask you to limit your verbal remarks to 5 minutes apiece, and I
would ask Mr. Sheaffer to go first.
STATEMENT OF JAMES W. SHEAFFER, PRESIDENT, NORTH AMERICAN
PUBLIC SECTOR, COMPUTER SCIENCES CORPORATION
Mr. Sheaffer. Thank you. Mr. Chairman, Ranking Member
Clarke, Mr. Duncan, it is an honor for me to appear before you
today.
My name is Jim Sheaffer. I am president of CSC's North
American Public Sector, with 29,000 employees who proudly serve
and support the missions of Federal agencies.
I also recently served as vice chair for the Public Sector
for TechAmerica Foundation's Commission on the Leadership
Opportunity in U.S. Deployment of the Cloud. In July our
commission issued a report called ``Cloud First, Cloud Fast''
that included 14 specific recommendations for the Federal
Government to accelerate the adoption of the cloud, and I
respectfully request that that document be entered into the
record.*
---------------------------------------------------------------------------
* The information has been retained in committee files.
---------------------------------------------------------------------------
Let me offer a brief word about CSC. Last year we had
revenues of just over $16 billion. We are acknowledged as a
leading global provider of I.T. services. We deliver large-
scale projects for both public and private sector clients, and
we provide cybersecurity to some of the world's largest
companies and some of the most sensitive U.S. Government
agencies.
By leveraging shared computing resources, higher
utilization rates of hardware, and economies of scale, cloud
computing is ushering in an I.T. revolution. Users pay only for
what they consume. Cloud computing and the as-a-service
delivery model enable organizations to cut costs of computing,
build capacity for growing volumes of data, and burgeoning
requirements for computation.
Cloud is a hot topic, but it is only the latest
evolutionary step in the field. I began first with custom-build
computers, moved to mainframes, on to personal computers, then
to client-servers, and then to the internet.
What is different about the cloud is the rate of adoption.
The economics are compelling and the take-up of this technology
is much faster than some of the earlier technologies that were
adopted. In fact, the global nature of the cloud makes this a
different kind of phenomenon.
Today's austere Federal budget climate offers an added
incentive for agencies to adopt the cloud, but it also raises
questions of trust. Trust is more than just security. U.S.
citizens and users must believe in the integrity and
reliability of cloud computing in addition to security.
We acknowledge the challenges. One, the speed of cloud
advancement requires new security policies and even new
security technologies and procedures.
The internet, which is the foundation for the cloud, was
originally designed without a primary focus on security, and
since then we have had to play catch-up to make it secure. In
the future it will require the design of intrinsically secure
architectures to ensure security.
A second risk is that all required security standards for
cloud are not yet in place, as we heard from the prior--the
previous panel. The National Institute of Standards and
Technology and the Cloud Security Alliance, a nonprofit
coalition, are developing, with industry support, those
standards, and we believe that they need to be global
standards, not just standards here in the United States.
Third, cyber threats are serious and dynamic, and becoming
more pernicious. Threats are more severe than we experienced in
the past, and the capabilities of bad actors are evolving
swiftly.
The risks and challenges to cloud computing are substantial
but not insurmountable and should not be used as an excuse to
shrink from the adoption of the cloud. Fundamentally,
cybersecurity must be integral to the architectures and not
bolted on after the fact. We at CSC are confident that prudent
cloud computing adoption can meet the stringent security
requirements.
How should those risks and challenges be addressed? The key
is to align the risk profiles of various types of data and
their uses with the levels of protection required.
One-size-fits-all approaches provide neither effective
security nor the lowest cost. Each application and data set
must be evaluated to identify its specific security
requirements, and then appropriate cloud solutions can be
implemented, choosing from private, public, or hybrid clouds.
As an evolving technology, it is important to gain feedback
and lessons learned from the implementation of cloud computing.
Lessons will need to be shared across agencies, as one of your
previous questions indicated.
The Department of Homeland Security is laudably reaching
out to foster a more secure and resilient cyber environment.
The Department is leaning forward to show leadership in cloud
adoption.
In consolidating infrastructure from the 22 components of
the primary data center at Stennis and its backup, DHS is
increasing the productivity of its capital investment in
computing and it has also implemented a private cloud behind
its firewall and security systems. The Department is clearly an
early and prudent adopter of cloud computing.
One example of the success of this approach is our systems.
With our assistants at DHS we are designing and implementing a
private cloud for DHS that will reduce the time to provision
new software development and test environments from months to
just a couple of days.
In conclusion, cloud computing offers enormous opportunity
to improve performance and reduce costs. Security issues can be
managed. The United States is a leader worldwide in cloud
adoption, and we can and must maintain that position.
I welcome your questions. Thank you.
[The prepared statement of Mr. Sheaffer follows:]
Prepared Statement of James W. Sheaffer
October 6, 2011
Mr. Chairman, Ranking Member Clarke, and Members of the
subcommittee, it is an honor to appear before you today to discuss
security implications of cloud--or shared--computing. The subcommittee
laid a good basis for today's discussion in its April 15 hearing on
promoting Department of Homeland Security cybersecurity innovation and
securing critical infrastructure, and its June 24 hearing on the
homeland security impact of the administration's cybersecurity
proposal.
I am Jim Sheaffer, President of CSC's North American Public Sector.
Recently I served as Vice-Chair for the Public Sector of the
TechAmerica Foundation's Commission on the Leadership Opportunity in
U.S. Deployment of the Cloud (CLOUD\2\). The mandate of the Commission
was to provide recommendations on how the Federal Government could
deploy and accelerate the adoption of cloud technologies, and to
address public policies that would enable U.S. innovation in the cloud.
In July, the Commission issued a report--Cloud First, Cloud Fast--that
addresses some of the issues we are discussing today.
Let me begin by offering a brief word about CSC. Last year we had
revenues of just over $16 billion. Three-fifths derived from IT
services provided to the private sector, and two-fifths from a range of
services for the public sector. Acknowledged as a leading global
provider of IT services, CSC delivers large-scale IT projects for both
public and private sector clients. We provide cybersecurity to some of
the world's largest companies, including critical infrastructure
providers, and some of the most sensitive U.S. Government agencies.
cloud computing
By leveraging shared computing resources, higher utilization rates
of computing hardware, and economies of scale, cloud computing is
ushering in an IT revolution which promises far lower costs while
greatly improving capacity and performance. Cloud computing combines
self-service provisioning of software applications and IT
infrastructure with on-demand scaling of computing and storage in which
users pay only for what they consume. Cloud computing and ``as-a-
service'' delivery enable organizations to slash unit costs of
computing, and build capacity for rapidly growing volumes of data and
burgeoning requirements for computation.
Cloud computing is a hot topic. In essence, it is just the latest
evolutionary step that has taken us from custom-built computers to
mainframes to personal computers to client-servers, and then to the
internet. What is different about cloud computing is the accelerating
pace of change, rapid adoption rates, and global nature of its use.
Cloud innovation allows entrepreneurs and public sector innovators
to create value at little to no capital expense in computing resources,
unlike the previous waves. Cloud computing disrupts existing business
models and enables wholly new ones. The explosion of mobile computing
catalyzes even faster adoption of cloud computing.
Cloud computing hardware can reside on-premise at an organization's
facility, or off-premise, such as at an IT provider's facility. The
National Institute of Standards and Technology (NIST) defines four
types of environments for cloud computing: (1) Private cloud that is
operated by an organization and may exist on-premise or off-premise;
(2) Community cloud that is shared by multiple organizations related to
a specific community and may exist on-premise or off-premise; (3)
Public cloud that is available to the general public, owned by a
commercial vendor and located off-premise; and (4) Hybrid cloud that is
a combination of two or more clouds (private, community, or public).
trust
Today's tight Federal budget climate offers an added incentive to
agencies to adopt the cloud. But while cloud computing offers
substantial benefits, such as cost savings, speed, and responsiveness
to mission needs, it also raises questions of trust. Trust encompasses
such concepts as security, availability, reliability, transparency to
the user, and ability to extract data.
The pace and degree of adoption of cloud delivery services will
depend on establishing a basis of trust. This begins with understanding
the risks and challenges. Can important data be entrusted to the cloud?
Are there new risks and challenges to trust, especially the security of
data?
Let us look at the new risks and challenges to trust. One, the
speed of cloud technology advancement requires new security policies,
and even new technologies and procedures, to keep pace with cloud
advancements. Most current knowledge about IT security is based on a
world in which most computer resources are under the direct control of
a person or organization and in which physical and technical means
exist, including software firewalls, to control access. Moreover, the
internet was originally designed without a primary focus on security;
since then computer security specialists have played catch-up.
Many of those security concepts must be reconsidered for a world in
which cloud computing enables a much broader spectrum of solutions and
much greater cost savings derived from the sharing of computing,
storage, and network resources, bringing new economies of scale. For
example, firewall technologies designed for operating inside the
virtual fabric of cloud architectures--the design of cloud computing
systems--are just now becoming available, and they remain largely
untested.
A second risk is that all of the required security standards for
cloud computing are not yet in place. Clear, understandable, and
verifiable standards are essential for building trust. The National
Institute of Standards and Technology and the Cloud Security Alliance--
a non-profit coalition of practitioners, companies, and associations--
are conducting research and developing new cloud security standards.
Third, while not specific to cloud computing but relevant to it,
cyber threats are serious and dynamic--and becoming more pernicious.
Business and Government alike face threats much more severe than in the
past, and more likely to change and do so swiftly.
Advanced Persistent Threats tend to be state-sponsored and target
especially sensitive information, such as military and financial data
and intellectual property. Such information lies at the heart of
America's security and economic well-being.
The risks and challenges to cloud computing are substantial but not
insurmountable. Of fundamental importance, cybersecurity must be
integral to cloud computing architectures and not be ``bolted-on''
after the fact. CSC participates in various forums that develop
standards. CSC's rigorous validation and testing programs promote
innovation for security solutions.
On balance, we are confident that prudent cloud computing will
satisfy stringent security requirements. USCYBERCOM Commander General
Keith Alexander said it best to a House Armed Services Subcommittee
last March:
``The idea is to reduce vulnerabilities inherent in the current
architecture and to exploit the advantages of cloud computing and thin-
client networks, moving the programs and the data that users need away
from the thousands of desktops we now use--up to a centralized
configuration that will give us wider availability of applications and
data combined with tighter control over accesses and vulnerabilities
and more timely mitigation of the latter.''
ways to enhance security
How should security risks and challenges be addressed? The key is
to align risk profiles of varying types of data and uses with levels of
protection required.
Understanding the risk profiles of data being considered for the
cloud is key to determining the required levels, and hence costs of
security. One-size-fits-all approaches provide neither effective
security nor the lowest-cost solution. Each software application and
data set must be evaluated to identify its specific security
requirements. For example, published scientific research may be
suitable for less-stringent cloud computing environments than are
needed for classified intelligence data on potential terrorists. CSC is
assisting Federal agencies to develop roadmaps that outline risk
profiles of data sets and identify appropriate cloud solutions.
It will be important to gain feedback and learn lessons from
implementations of cloud computing. They can help identify best
practices and improve security for future uses.
federal policy
Federal policy on cloud computing and its security has evolved
rapidly. In 2002 the Federal Information Security Management Act, or
FISMA, came into force. It establishes a ``comprehensive framework
designed to protect government information, operations and assets
against natural and man-made threats,'' and requires program officials,
chief information officers, and inspectors general to conduct annual
reviews of information security.
The Federal Risk and Authorization Management Program, or FedRAMP,
was initiated in 2010 to provide a standard approach across the Federal
Government for assessing and authorizing cloud computing services and
products. A common security risk model enables the Federal Government
to ``approve once, and use often.''
In the 25-Point Implementation Plan to Reform Federal Information
Technology Management, issued on December 9, 2010, the Office of
Management and Budget called for reducing the number of Federal data
centers by at least 800 by 2015 and creating a Federal-wide marketplace
for data center availability. Curiously, not one of OMB's 25 points
focused on cybersecurity.
On February 9, 2011, OMB issued a Federal Cloud Computing Strategy,
which gives more attention to security. It cautions that cloud security
is an exercise in risk management, ``identifying and assessing risk,
and taking the steps to reduce it to an acceptable level.'' Risk
management based on intelligent risk assessment enhances the protection
of the most valuable information and is more cost-effective than
compliance-based approaches.
The Federal Strategy points to several potential security benefits
of cloud computing. The first is the ability of the cloud provider to
focus centralized resources on security services. Second, the greater
uniformity and homogeneity of the cloud platform eases security
management and improves response times. A third benefit is the improved
resource availability of the cloud provider through scalability,
redundancy, and disaster recovery capability. Fourth are the improved
backup and recovery capabilities and procedures that a cloud provider
can offer. A fifth potential benefit of cloud computing is the ability
to leverage, as needed, services from other data centers.
At the same time, the Federal Strategy highlights potential
vulnerabilities of cloud computing. One is the inherent system
complexity of a cloud computing environment. A second vulnerability is
dependency on the service provider to maintain secure logical
separation in a shared computing resource, or what is called a multi-
tenant environment. A third potential vulnerability is the cloud user's
need to have sufficient knowledge of potential threats and
vulnerabilities to know how to make decisions and set priorities on
security and privacy.
Increasing experience in the implementation of cloud computing,
with careful attention to security, will help validate and refine our
collective understanding of its benefits and risks.
The Department of Homeland Security is laudably reaching out across
the Federal Government and the private sector to foster a more secure
and resilient cybersecurity environment. The DHS Chief information
Officer is leaning forward to show leadership in cloud adoption.
In moving data from 22 separate components into the primary DHS
Stennis data center and a secondary backup center, DHS has increased
the productivity of its capital investment in computing. While
migrating into the two consolidated data centers, DHS has also
implemented a private cloud behind a DHS-controlled firewall and
security systems. As new security standards are developed and
effectively verified, more data will be ready to move to the cloud. In
addition to private cloud implementation, DHS is moving certain public-
facing websites, such as DHS.gov and FEMA.gov, into a public cloud in
order to increase efficiency and productivity. DHS is an early and
prudent adopter of cloud computing and its experience may be
instructive for others.
cloud examples
Let me outline three examples of how cloud computing can be
implemented in a homeland security context.
First, CSC helps a global chemical company that is part of
America's critical infrastructure. Its research unit must allow access
to scientists and others from inside and outside the company to foster
collaboration for new discoveries. Researchers require high-performance
computing and surge IT capacity, and they store highly sensitive
intellectual property. The research unit must accommodate projects that
start and stop abruptly and then restart.
CSC has installed a private cloud that the chemical company manages
to satisfy its own special security requirements. The company has
deployed cloud access at each of its laboratories around the world, and
CSC federates and orchestrates cloud services across the chemical
company's global IT infrastructure.
In a second example, DHS wanted more responsive computing. It opted
for cloud computing for the development and testing of new computer
application systems. This eliminates costly and time-consuming tasks of
procuring, installing, and testing new computer hardware and software
every time a software development team starts a new project.
To support DHS, CSC designed and is implementing a private cloud
that will reduce the time to provision new development and test
environments from months to just a couple of days. We are also
assisting with a strategy and plan for helping DHS encourage management
and cultural changes required to take best advantage of the cloud.
A third example is the potential for increased use of unmanned
aerial vehicles to help DHS monitor U.S. borders. Evolving technology
will allow aerial platforms to collect greatly increasing amounts of
ground imagery. As this develops, cloud computing could assist DHS to
expand data collection and processing while holding down computing
costs.
recommendations
I wish to call special attention to four important recommendations
from the TechAmerica Commission Report, and offer a fifth
recommendation.
First, the Federal Government and the private sector should support
the creation of international standardized frameworks for securing,
assessing, certifying, and accrediting cloud computing.
Second, the public sector and the Federal Government should
accelerate the development of an identify management ecosystem to
facilitate the adoption of strong authentication technologies, enabling
more secure access to cloud services and websites.
Third, a law is needed to clarify responsibilities of companies to
notify customers in the event of data breaches, and strengthened
criminal laws are required against those who attack computer systems,
including cloud services.
Fourth, the Federal Government and the private sector should
develop and execute a more robust joint research agenda for cloud
computing.
Fifth, verification and continuous monitoring of cloud security
ought to be standardized. Independent, professional third-party audit
of cloud providers should become standard practice, along with real-
time transparency in the security posture of cloud-based systems.
conclusion
In conclusion, as the use of cloud computing accelerates, better
security must go hand-in-hand with saving money and improving
performance. Cybersecurity must be integrated into cloud computing
architectures at the outset, rather than be left to ``catch up.'' This
will enhance trust in the information revolution that underlies so much
of America's prosperity and homeland security.
I welcome your questions and comments. Thank you.
Mr. Lungren. Thank you very much, Mr. Sheaffer.
Now, Mr. Brown.
STATEMENT OF TIMOTHY BROWN, SENIOR VICE PRESIDENT AND CHIEF
ARCHITECT FOR SECURITY, CA TECHNOLOGIES
Mr. Brown. Chairman Lungren and Members of the
subcommittee, I want to thank you for the opportunity to talk
to you today. CA Technologies is one of the world-leading I.T.
management software companies that provides software and
services to enterprise, governments, and cloud providers.
The hype and promise to the cloud continue to accelerate,
but it is clear that significant confusion remains about
exactly what cloud computing is and the risks and benefits
associated with it. Security is the concern cited most.
When you consider the loss of direct control involved with
the cloud these concerns are expected, but they must be
addressed for the cloud to be successful. From a security
perspective, any service that is accessed outside of an
enterprise's direct control should be considered a cloud
service.
Services like ADP, for check processing, and a 401(k)
portal are good examples of--that have been around for a long
time. Cloud is not new, but the current momentum and explosion
of new cloud services gives us opportunity to enhance
cybersecurity.
Mr. Lungren. I think we lost your mike there.
Mr. Brown. Am I back?
Mr. Lungren. There you are.
Mr. Brown. All right. We will move up. Here we go.
So CA Technologies believes the responsibility for cloud
security lies with both the providers and the consumers. The
cloud is neither inherently more secure nor less secure than
other I.T. services.
Security fears and arguments that those fears are overblown
have muddied the waters about this vital issue. To provide some
clarity I will focus on four critical areas affecting cloud
security.
First, it is important to note that cloud won't replace all
other technologies and service delivery options. As
organizations move to the cloud it will be one of many
platforms that must be operated and managed together to
minimize risk and security vulnerabilities. We should be wary
when people say that cloud will replace all technologies.
Second, the responsibility for security rests with both the
provider and the consumer of cloud technologies. Different
cloud services have different risk profiles.
What is important is transparency. Customers and providers
need to agree upon those security expectations and know that
the service being deployed meets those requirements.
Customers must have trust in their cloud service providers
but also must have the ability to verify their claims and
performance. Cloud customers need to be vigilant in their
investigation, auditing, and oversight of their providers.
Cloud providers must approach securing their customers' data
with the same degree of seriousness as the owner of the data.
Third is that a strong trusted identity system that enables
the right people to have the right access to the right
information at the right time is vital to securing the cloud.
Many of the data breaches we read about today find their root
cause in weak identity and access management controls.
To be certain the move to the cloud doesn't create new
security risks cloud consumers should ask the following: Who
has and needs access to what? What can they do with that
access? What can they do with the information they obtain?
Finally, what did they do with that information?
On-line banking and bill pay services provide an example of
how transactions between different cloud services can be
accomplished using strong identity management. As most of us
know, different on-line banking transactions have different
risks, and banks have implemented tiered security requirements
based on that risk. Simply accessing your account balances
requires one level of authentication, while transferring funds
may require a higher degree of security.
If you want to authorize your bank to pay a bill, your bank
may need to access a bill payment service in the cloud on your
behalf. This type of transaction requires that the bank and the
bill pay service have trusted and transparent security
practices that are audited and enforced.
Finally, the adoption of standards is critical to the
security and operability in the cloud. CA Technologies
contributes actively to the efforts of standards organizations,
such as OASIS, and collaborates with NIST.
There are two efforts I would like to highlight. The first
is FedRAMP.
FedRAMP offers the promise that solutions can be accredited
once and used many times across Federal agencies. While we
await the final draft of FedRAMP, several questions about its
scope and its implementation remain. We recommend that Congress
continue oversight to be sure these important questions are
answered.
The second is the National Strategy for Trusted Identities
in Cyberspace, or NSTIC. NSTIC is aimed at enhancing trust by
strengthening industry-based identity management practices and
minimizing the proliferation of username and password
combinations we use on-line.
NSTIC has asked for its first budget in fiscal year 2012.
We recommend that Congress fund this important effort.
Finally, I would like to offer several additional
recommendations for your consideration. First, because we are
in the nascent stage of cloud adoption, Congress should look at
cloud policy issues through the lens of outcomes, not specific
technologies. Static rules and mandated checklists are not
adequately flexible and will rapidly become outdated as new
technologies emerge.
Second, Congress should avoid adopting policies that create
a country-specific--country-specific policy. For U.S.
businesses in competing markets all over the world, global
harmonization policy will enable industry to build solutions
that can be delivered in multiple markets and will enhance our
competitiveness.
Finally, the cloud is an opportunity for new business
models, enhanced security, and for the United States to drive
innovation and technical leadership. We recommend that Congress
support the important policy recommendations from the
TechAmerica Cloud\2\ commission.
I appreciate your opportunity to be here for you today. I
would be happy to answer any questions. Thank you.
[The statement of Mr. Brown follows:]
Prepared Statement of Timothy Brown
October 6, 2011
Good morning Chairman Lungren, Ranking Member Clarke, and Members
of the subcommittee. My name is Tim Brown and I'm honored to be here
today to testify on cloud computing security risks and opportunities. I
am the senior vice president and chief architect for security at CA
Technologies. CA Technologies (www.ca.com) is one of the world's
largest information technology management software providers. The
company has expertise across IT environments--from the mainframe and
distributed computing to virtual and cloud technologies. CA
Technologies manages and secures IT environments and enables customers
to deliver more flexible IT services. The majority of the global
Fortune 500 and most major Federal and State government agencies rely
extensively on CA Technologies software to manage their constantly
evolving technology environments. Founded in 1976, CA Technologies is a
global company with headquarters in New York, 150 offices in more than
47 countries, and thousands of developers and researchers worldwide.
CA Technologies was honored to serve on the TechAmerica
Foundation's Commission on the Leadership Opportunity in U.S.
Deployment of the Cloud (CLOUD\2\), and was heavily involved in the
development of the Commission's recommendations. Since another member
of the Commission is participating in the hearing today, I will focus
the bulk of my remarks on a number of specific cloud security issues CA
Technologies believes are critical to ensure secure adoption of cloud
computing. However, CA Technologies supports the recommendations of the
CLOUD\2\ report and I address many of the issues covered in the
Commission's report in my testimony today.
CA Technologies believes that cloud computing is neither inherently
more nor less secure than other IT platforms, and that securing the
cloud is a shared responsibility of both providers and consumers of
cloud services. There are a number of policy issues that must be
resolved to realize the cloud's potential and we will focus on those
issues on our testimony today.
introduction
While both the hype and promise surrounding cloud computing
continue to accelerate at a feverish rate, it is clear that significant
confusion remains in global markets about what exactly cloud computing
is and what the risks and benefits are associated with transitioning to
this latest technology. Corporate and governmental organizations across
the globe are anxious to reap the cost, performance, and agility
benefits that the cloud can offer, but at the same time are wary of a
range of risks that accompany a different way of buying and consuming
technology solutions.
Chief among concerns raised in survey after survey of both current
and potential cloud customers is security. Security is often followed
by related concerns about data privacy as well as interoperability,
availability of cloud services, performance, and transparency of
providers. When one considers the loss of direct control that
accompanies cloud deployments, concerns about security risks associated
with moving to the cloud are not only reasonable, but also expose
critical operational risk management issues that must be discussed and
addressed when determining if and when to move particular services to
the cloud.
It is important to keep in mind that from a security professional's
perspective, any service that runs outside of the operationally-
controlled environment of an IT organization is considered a cloud
service. This is true in the case of commonly-known cloud services like
Salesforce.com, Google Docs, and cloud email, but also includes
services like ADP, 401(k) programs, corporate travel sites, and health
plans. No two applications or systems are alike, and pragmatic
implementation of cloud technologies necessitates that risk-based
processes be used to determine what services and applications may or
may not be feasible to move to the cloud, their level of sensitivity,
what platform is most suitable, whether a private or public cloud
environment is appropriate, and the specific security and operational
controls that are needed.
The use of cloud computing represents an exciting new opportunity
for IT organizations and for CIO's in both business and Government to
remake the way in which they work together with their customers and the
user communities that rely on IT-based services. Because cloud
computing enables IT organizations to focus on business services rather
than infrastructure, technology organizations will have increased
agility to build new solutions to support their customers with minimal
investment.
In my testimony today, I would like to focus on the four key areas
that CA Technologies feels must be considered in evaluating both the
opportunities and risks associated with the transition to cloud:
The reality of new complexities introduced with the adoption
of cloud computing;
Security considerations for the cloud;
The critical role that identity management and
authentication play in enabling cloud security; and
The importance of standards development and adoption to
ensure interoperability and common implementation of cloud
solutions globally.
I will also make some recommendations on the role Congress can play
in fostering the secure uptake and adoption of cloud computing
solutions.
the ``new normal'' of cloud computing
A theme that CA Technologies keeps hearing from our customers is
that they want to use cloud computing as a real game-changer. The
layers and layers of complexity in IT have made it increasingly more
challenging to deliver new services to the business in a rapid manner.
The global downturn in markets across the globe has resulted in flat
and/or declining IT budgets in both the commercial and public sectors.
But the demand for new technology-based services inside large
organizations has not slowed, so IT organizations are constantly
challenged to provide more business technologies faster with reduced
resources.
These factors have all contributed to the perfect storm that has
emerged for cloud uptake across the globe.
It is important to note that while many would have you believe that
cloud technologies will replace all on-premise IT, in reality the
transition to cloud technologies will be gradual and the need to
develop and support on-premise solutions will remain for the
foreseeable future. The introduction of cloud technologies will create
greater complexities for IT organizations to manage and support. With
cloud solutions, a single business service may include a combination of
physical, virtual, and cloud components that all must work together to
deliver the functionality that users expect.
Consumers of cloud technologies will find themselves in a hybrid
technology environment for a long time. Existing solutions and
technologies will still need to be maintained, and cloud technologies
will most often serve as a natural extension of existing IT
environments. As such, the cloud introduce a new heterogeneity to IT
environments, one that will require coordinated and orchestrated
management, transition plans, and risk-based security evaluations.
This can be a real boon to IT organizations that can harness the
enthusiasm and momentum of the cloud to drive changes that have been
needed in the management process for technology generally. One of the
most promising aspects of cloud computing is its ability to fill the
gap between technology supply and demand and help organizations focus
less on commodity IT services and more on what is unique to their
particular business or Government program. Off-loading standard
services and functions to the cloud can save money and resources that
can be better utilized to drive change and tackle problems that are
more foundational and transformative to businesses and governments. At
CA Technologies, we call this opportunity the innovation dividend.
To gain this dividend, however, IT organizations must take a very
focused and methodical approach to evaluating what should or should not
be moved to the cloud. The means that organizations need to evaluate
people, processes, technology, and perhaps most importantly, risk
involved with each potential opportunity move to the cloud.
Organizations may determine that certain services, applications and
data are too critical or sensitive to be moved to the cloud, which can
be an appropriate risk management decision. The cloud is not a panacea,
and may not be appropriate for all workloads. Organizations must take a
measured approach that is driven by substantive analysis of the risks
and opportunities associated with each opportunity to migrate services
to the cloud.
Once decisions have been made to move a particular service or
application to the cloud, organizations must evaluate what providers
and what services will meet their needs. All of these analyses have
impacts on and contribute to the security posture of the organization.
Some of the considerations that CA Technologies advises our customers
to use in evaluating providers include the following, which have been
developed through the Cloud Service Measurement Initiative Consortium
(CSMIC) that I provide additional details on later in my testimony:
Accountability.--Can we count on the provider to deliver the
promised service?
Agility.--Can the service be changed, and how quickly?
Assurance.--How likely is it that the service will work as
expected?
Cost.--How much is it, including both start-up and on-going
costs?
Performance.--Does the service do what we need?
Usability.--Is it easy to learn and use?
Portability.--Can I move my data and application from one
provider to another?
Security and Privacy.--Is the service safe and privacy-
protected?
security issues in the cloud
Just like when you buy a car, an appliance, or any other service,
the reputation of cloud providers and their ability to deliver on the
service promised is a key consideration when making a purchase of cloud
solutions. The Cloud Service Provider ecosystem is just as diverse as
any other industry. Responsible providers want to do all they can to
demonstrate trust and accountability to their customers and that
security services are built in and not bolted onto their solutions.
These providers will be in the cloud marketplace for the long run and
will continue to drive innovation and excellence in the industry. But
it is important to keep in mind that new and innovative cloud service
providers are emerging daily. We are in the midst of a significant
expansion period in the cloud market, and the ever-expanding number of
providers who want to move into the cloud market may not have long-term
interest or commitment to the technology, which in turn may create
risks for customers who want to embrace the cloud. Customers must have
assurance their provider of choice will be there when they need service
modifications or need to move their data and applications elsewhere,
and that they take the responsibility of securing their data as
seriously as they do as the owner of that data.
The Cloud Security Alliance (CSA), a major industry consortium
focused on cloud security issues, has identified 14 critical focus
areas for organizations deploying cloud computing resources.\1\ CA
Technologies/Ponemon Institute survey of the cloud service provider
community made use of these 14 areas in a report released earlier this
year. The survey data uncovered a wide range of viewpoints on the role
that cloud service providers have in providing security for their
solutions. With lower costs and faster deployment being the main
drivers for moving to cloud services, some providers feel that security
is more the responsibility of cloud customers than it is of providers.
---------------------------------------------------------------------------
\1\ The 14 focus areas identified by the Cloud Security Alliance
are the following: Governance and enterprise risk management; legal and
contracting issues; procedures for electronic discovery; compliance and
audit; information life-cycle management; portability and
interoperability; business continuity and disaster recovery; data
center operations; incident response, notification, and remediation;
application security; encryption and key management; identity and
access management; storage operations; and virtualization operations.
---------------------------------------------------------------------------
In reality, not all cloud services require the same level of
security. It will be appropriate for certain workloads to be deployed
in the cloud with different security levels than others. But the goal
of cost savings that is so often identified as the main driver for
cloud adoption sometimes masks the importance of security risk
management. Security must remain at the forefront of all cloud strategy
discussions to ensure the right sets of security controls are applied
to the right services. What is important is that security, performance,
cost, and accountability decisions are clear and transparent to the
users of cloud services.
CA Technologies believes that the responsibility for securing the
cloud lies with both the providers and the consumers of cloud
solutions. The cloud is neither inherently more nor less secure than
other IT services and solutions. Generalized concerns over cloud
security on the one hand, and arguments that the security risks in the
cloud are overblown on the other hand, have muddied the waters to the
point that policymakers and practitioners are experiencing security
schizophrenia. Should I overlook legitimate security concerns and
plunge head-first into the cloud, or should fear and uncertainty of
these risks stop me from doing anything that even remotely resembles
cloud computing? Like most responsible decisions, the answer lies
somewhere in the middle of these two extremes.
Cyber criminals, state and non-state actors, and other cyber
adversaries move rapidly and adeptly to exploit weaknesses and
vulnerabilities in systems, networks, applications, and practices. They
are successful at taking control of machines and stealing data. But
done right, the movement to the cloud is an opportunity for
organizations to enhance operational security.
As such, potential consumers of cloud solutions must be mindful of
the wide range of providers and the security risk management controls
they have implemented for the solutions they host or provide in the
cloud. A key for cloud customers will be to evaluate both the
sensitivities of the services and data they hope to deploy to the
cloud, and a long-term viability, references, and the depth of their
solutions.
Cloud customers must insist on built-in security and transparency
from the providers they select. They need to create compliance plans
and closely scrutinize their contracts, Service Level Agreements
(SLAs), and the security and disaster recovery plans of their providers
to ensure they are making sound choices on who to partner with in
moving services to the cloud. A key consideration here is to trust, but
verify. CA Technologies recommends that cloud customers meet their
responsibility to audit and monitor their providers, including the use
of inspection programs, testing and monitoring compliance with SLAs,
and assessing the security of critical systems.
identity and access management as a foundation of cloud security
While there are certainly myriad operational issues to consider
when architecting cloud solutions to deliver strong and robust
security, CA Technologies believes that identity and access management
(IAM) issues deserve particular attention. Our surveys of cloud
providers and the views from leading industry analysts and
organizations find that identity and access management is the most
important issue that companies considering moving to the cloud face
today. A strong trusted identity system that enables the right people
to have the right access to the right information is critical to the
protection and enablement of the cloud.
Cloud service providers and customers generally feel comfortable
that they have highly qualified IT personnel and tools which can
prevent or curtail viruses from infecting their services, and that they
can effectively secure data flowing in and out of cloud services. They
are less comfortable with the process of identifying and authenticating
the users, systems, and devices that need access to their services and
managing access to specific information or data when using cloud
services.
One of the greatest challenges facing the IT sector today is
fostering on-line trust, including the important trust components of
security and privacy. The fact is that most on-line threats and
successful data breaches of late have been based on and exploit access
control and identity management failures in systems. The Government
Accountability Office has written to Congress about unauthorized access
issues as recently as Monday of this week (October 3, 2011). Identity
management and access management controls are central to the secure
adoption of cloud services.
Identity and access management practices within the cloud provide
the foundation for effective security by ensuring that all users have
only the appropriate level of access rights to protected resources, and
that those rights are effectively enforced. IT organizations generally
as well as cloud service providers, both public and private, struggle
to keep up with the explosion in the number of users from multiple
systems, applications, and user communities that are consuming their
services and the complexity of managing access rights for these users.
With the transition to cloud solutions, employees and applications
will continue to move outside the walls of the customer enterprise.
This introduces new risks for unauthorized access and the loss of
information. Cloud applications are new services that users must have
access to, and managing that access without creating new
vulnerabilities or new silos of identity are incredibly daunting
challenges. Managing the on-boarding and off-boarding of users to cloud
services and integrating those access rights with the overall IAM
strategy for on-premise solutions requires that cloud providers and
customers answer the following questions:
Who has and needs access to what?
What can they do with that access?
What can they do with the information they obtain?
What did they do with that information?
These questions reinforce that managing access and authorization is
but one part of the challenge. To be successful, identity security
strategies must also focus on the specific data being accessed and what
individual users can do with it. CA Technologies refers to this process
and approach as content-aware identity and access management.
Cloud computing creates opportunities for Government agencies and
commercial organizations alike to make certain that new silos of
identity don't emerge that increase vulnerabilities and complexities
for users. For Government programs and systems, we recommend that
Federal agencies enhance their IAM capabilities to provide for risk-
based authentication, the use of multi-factor authentication solutions,
and leverage the investments they have already made in Personal
Identity Verification (PIV) cards.
An example of how many of these integrated identity controls are
used today can be found in the financial services sector. CA
Technologies counts the majority of the world's major financial
services organizations as customers, and we have worked closely with
these organizations to implement strong and flexible IAM solutions that
provide their customers with ease of use in the most secure fashion
possible. Financial services firms have taken a security-first approach
because of the economic risks of the transactions they conduct.
Enhancing the security of those transactions helps meet regulatory
requirements, but first and foremost focuses on providing Defense in
Depth in ways that enhance security and provide ease of use for
consumers that include IAM solutions as a core component. Financial
institutions are doing a great job of analyzing not only the risk of
individuals and their access rights, but also the unique risks of
individual transactions. This is a trend that we believe the overall
cloud security market must and will embrace.
Most of us are already comfortable with the concept of signing onto
the website of our bank to access our account information. This usually
requires that users provide an account number, username, and password.
If you want to move money around from one bank account to another at
the same financial institution, the bank may require you to provide a
secondary identifier, like a PIN, because that transaction involves
more risk. If you want to use your bank's bill pay service and
authorize the movement of money from your bank to your credit card
company or your local utility, the transaction becomes more complicated
and introduces additional risk to both parties involved.
In many cases, when you initiate a transaction like this from your
bank, the experience to the user will be seamless. But behind the
scenes a complex transaction whereby the user is redirected to a bill
pay website and has their identity credentials passed to the bill pay
provider without needing to sign on or provide their credentials again
has taken place securely and transparently. The identity authentication
taking place in this scenario is being accomplished via a cloud
service. This type of transaction is an illustration of how user
experience and sound security can be implemented across the very
diverse technology environments present today. We believe that this
represents the direction future secure transactions across public,
private, and hybrid cloud environments will progress.
the role and need for standards in fostering cloud security
I believe this example also highlights the importance of standards
development and the valuable contributions of industry-led, recognized
standards development organizations (SDOs) and consortia. The adoption
of standards and their integration into the innovative security
solutions offered by the vendor community make possible predictable,
interoperable, secure implementations in enterprise and cloud-based
services. Such standards are vital to the management of cloud security
risks. As I noted earlier, existing security technologies implemented
in the enterprise are the building blocks of cloud security. And to a
huge extent those technologies, and the practices and controls which
they support, are standards-based.
Such building block standards are now foundational for cloud
computing environments, and where gaps exist, new standards are under
development. CA Technologies and other major IT companies contribute
actively to these efforts. For example, the Organization for the
Advancement of Structured Information Standards (OASIS) has developed
important security standards such as Extensible Access Control Markup
Language (XACML), Security Assertion Markup Language (SAML), and web
services security standards such as WS-Trust. OASIS also has technical
committees in place addressing new security challenges applicable to
the cloud, such as cloud identity, identity trust elevation, privacy
management, and reputation management. Its committees are also working
to create profiles which are used to apply existing standards suchas
XACML directly in support of cloud computing requirements.
Other standards bodies, including the Internet Engineering Task
Force (IETF) and the World Wide Web Consortium (W3C), de jure bodies
such as the International Organization for Standardization/
International Electrotechnical Commission Joint Technical Committee 1
(ISO/IEC JTC 1), key industry consortia such as the Open Identity
Exchange and the Kantara Initiative and other standards organizations
are all key contributors to enabling trust in the cloud. In combination
with best practices organizations such as the Cloud Security Alliance,
the resources contributed by industry, academia, governments, and
independent technical experts together represents a huge and on-going
investment to support security risk management in the cloud
environment. I would like to note the important role that the National
Institute for Standards and Technology (NIST) plays by its active
participation in industry standards development and as a convener of
industry efforts and focus. NIST recently issued a Special Publication
500-291, the Cloud Computing Standards Roadmap, which examines the
applicability of standards for the cloud and areas where gaps need to
be filled.
The NIST publication looks well beyond security alone, and SDOs and
consortia have certainly recognized the importance of standards-based
cloud interoperability at the data level, and through the development
of relevant application, operational management, license management,
audit, virtualization, and other standards that are needed to enable
interoperability of applications and services across clouds. CA
Technologies is a major participant and leader at many levels of the
cloud standardization process. And we believe that all of these
categories of standardization, and more, are relevant to the
development of interoperable clouds and cloud computing trust.
There are several specific efforts I want to highlight as examples
of emerging standards in the cloud security arena. The first and
perhaps most important in the Federal space is the Federal Risk and
Authorization Management Program (FedRAMP). While still in its draft
form, FedRAMP will provide Federal agencies with a baseline, common
approach for assessing and authorizing cloud services for use in
Federal agencies. This will provide Federal agencies with a common set
of controls against which to evaluate cloud services, and will give
cloud providers certainty of Federal specifications that must be built
into their products. FedRAMP is built on the premise that solutions
should be certified once and used many times across Federal agencies.
Federal agencies, however, have shown a tendency historically to ignore
previous certifications and re-certify technologies for use in their
own departments based on special requirements. Reciprocity of
authorizations will be a critical gauge of the success of FedRAMP.
FedRAMP will also require the transmittal of more frequent
operational security information by providers to the Government, a
process that is most-often termed ``continuous monitoring.'' Continuous
monitoring offers the potential to dramatically improve the situational
security posture of Federal information systems that rely on the cloud
if implemented correctly.
While we await the final draft of the FedRAMP specifications,
several questions about its scope and implementation remain, however.
Will agencies be required to honor authorizations made by other
agencies and avoid re-evaluating solutions that are implemented
similarly at another agency? How often and how will the security data
envisioned under continuous monitoring be transmitted? How will the
Government evaluate this data once received? The answers to these and
other questions will be critical to ensuring FedRAMP is both
implemented correctly and receives the buy-in needed from Government
and the private sector to ensure its success.
A second area I feel is important is the need to develop common
service measurement frameworks to help enable data-driven decisions on
the relative effectiveness of cloud solutions based on variables like
cost, availability, security, and scalability. Right now, there is no
standard mechanism to evaluate common services from different providers
against one other. The Cloud Service Measurement Initiative Consortium
(CSMIC), under the direction of Carnegie Mellon University and with
participation from government agencies like the State of Colorado
Office of the CIO, and corporations like CA Technologies and Accenture,
has begun developing a service measurement index (SMI), which can be
used to measure and compare a business service using a common language
and evaluation process. A high-level representation of the
characteristics and questions the CSMIC seeks to address is included as
an attachment to my testimony today. In conjunction with standard
recognition of cloud services authorized under the FedRAMP program, the
use of a framework like SMI in Government procurements will enhance the
analysis of competing cloud services and lead to greater
standardization of solutions. As such, CA Technologies encourages the
U.S. Government to investigate using the SMI to encourage data-driven
decision-making on cloud acquisitions.
Third, in the area of identity and access management, the National
Strategy for Trusted Identities in Cyberspace (NSTIC) is a critical
initiative that will make it easier for citizens and consumers to
securely and confidently navigate cyberspace and will enhance trust
among different consumers of identity through the sharing and
reciprocation of identity credentials. NSTIC is aimed at enhancing on-
line trust by strengthening industry-based identity management
practices and minimizing the constant proliferation of username and
password combinations that individuals must remember to conduct
business on-line. The standards and governance rules that will be
developed under NSTIC are a critical component of implementing robust
IAM solutions that can enhance trust of and the use of cloud computing
services. As the NSTIC program gets up and running at the Department of
Commerce, CA Technologies recommends that Congress fully fund this
important effort and that Federal agencies become active participants
in both the development of the NSTIC standards, and ultimately, accept
private sector-issued credentials as a means of authentication for
citizens who wish to interact with Government agencies securely.
Standards development, then, is an on-going and vital area of
industry and Governmental focus. It is international in scope, and the
standards are integral to key Government initiatives such as FedRAMP
and NSTIC. It is important that the subcommittee recognize that it is
only through support for industry-led, internationally supported
standards will we have measurable, interoperable security risk
management technologies, innovative technical solutions and practices
that can ensure trust in cloud-based services, not only in the United
States, but globally.
recommendations for congress
I was asked to address some of the security risks and opportunities
associated with the transition to cloud computing. I hope that my
testimony has highlighted that while there certainly are risks, the
opportunities are extremely positive if a number of actions are carried
out to ensure that the adoption of cloud technologies does not create
new silos in IT security and new, unintended risks. We are in the
nascent stage of cloud adoption. To ensure the promises of cloud
computing can be delivered in concert with effective security risk
management, we recommend that Congress:
Adopt policies that can accommodate future development and
flexibility in the cloud market, specifically, and in IT more
generally. Too often, Federal policy has imposed static
frameworks that must constantly be updated based on new
technology developments. We recommend that Congress focus on
outcomes and not on specific technologies;
Avoid policies that create a fragmented, country-specific
market for cloud services in the United States. As the cloud
market continues to evolve, we see great risk for market
segmentation based on unique policies designed solely to
address U.S. or other countries' market demands. For U.S.-based
businesses seeking to compete in markets all over the world,
globally harmonized policies will enable industry to build
solutions that can be delivered in multiple markets, enhances
our competitiveness, and makes it easier to deliver innovative
solutions around the world. Policies that acknowledge the
global nature of cloud markets will enable the United States to
maintain its leadership position in cloud computing and
encourage innovation to support jobs and exports of U.S.-
developed technologies;
Support standards developed by recognized standards
development organizations in the areas of cloud security,
interoperability, and transparency. These standards are vital
to the management of cloud security risks and should be
embraced by Congressional and Executive Branch policy makers;
Fund and support the continued development and rollout of
FedRAMP and the NSTIC. To enhance operational cybersecurity at
the Federal level, we recommend that Congress ensure that
critical funding to develop and implement these programs be
preserved, even in difficult Federal budget environments. We
further recommend that Congress keep a watchful eye on FedRAMP
implementation to ensure that the efficiencies hoped for are
achieved;
Continue support for NIST and its unique role as both an
internationally-respected body of security experts developing
standards and practices for the Federal Government as well as
for its important function as a contributor to industry-led
standards development and as a convener for addressing emerging
security issues; and
Encourage the Federal Government to leverage emerging
efforts to develop service measurement indexes in Government
cloud procurements. The CSMIC effort I described in my
testimony can provide Federal agencies facing budget,
performance, and transparency demands with tools that take
data-driven approaches to evaluating competing offers of cloud
technologies. I believe that frameworks like these can
facilitate more robust decision-making about which specific
cloud services and providers are right for Federal agencies.
Mr. Chairman, Ranking Member Clarke, and Members of the
subcommittee, this concludes my written statement. I appreciate the
opportunity to appear before you to share some of our thoughts on cloud
security. CA Technologies shares the subcommittee's goal of increasing
awareness of the cloud and the particular goal of enhancing
cybersecurity, and we would be happy to work with you towards this goal
however we can.
I would be happy to answer any questions you may have for me.
Thank you.
Mr. Lungren. Thank you very much.
Now, Mr. Bottum, you are recognized for 5 minutes.
STATEMENT OF JAMES R. BOTTUM, VICE PROVOST FOR COMPUTING AND
INFORMATION TECHNOLOGY AND CHIEF INFORMATION OFFICER, CLEMSON
UNIVERSITY
Mr. Bottum. Mr. Chairman, I would like to thank you and the
Members of the subcommittee for the opportunity to present this
testimony. Located in Clemson, South Carolina, Clemson
University is a Nationally-ranked public land grant research
university with an enrollment of 19,500 students.
Mr. Chairman, many definitions explain what the cloud
represents. A good working definition should reflect the
distinctive characteristic of cloud computing, namely on-demand
delivery of shared services over the internet.
By allowing users to share resources, cloud computing
enables infrastructure to be right-sized, balancing user
requirements with the information technology services actually
rendered. Cloud computing is both efficient and economical.
However, we must ensure that our security tools, practices, and
policies grow in proportion to our use of this evolving
technology.
Clemson has, in some sense, been in the cloud business for
over 30 years, provisioning Medicaid applications and services
to the State and citizens of South Carolina. Three years ago,
as the recession intensified, we created a South Carolina Cloud
experiment to see if several institutions could do things we
could not do by ourselves, and/or do them in a more economical
fashion.
Today our cloud is operational and involves a collaboration
of educational institutions and commercial organizations.
Partner institutions include both public and private
universities, technical colleges, and historically black
colleges and universities. Many of these would not ordinarily
have access to the resources as a stand-alone institution.
Our team is working with a Fortune 500 company to build out
a secure and comprehensive cloud computing environment.
Considering our diverse set of users and the numerous
organizations that connect into the environment, it is
important to properly ensure identity and access management and
address concerns over data theft or manipulation and
vulnerabilities.
Our goal is to apply policies, procedures, and controls
that are seamless, transparent, and non-impeding to the end-
user. It is my view that the benefits of cloud computing far
outweigh the risks.
A thoughtful strategy for prudently broadening adoption of
cloud services can facilitate a smooth transition to this
dynamic platform. The transition should be complemented with a
thoughtful and comprehensive information security initiative to
ensure the protection of our data and resources as our
environments have evolved.
To increase security within the cloud, R&D is needed in a
number of areas. Six important areas are highlighted here.
The first area involves the use of virtual machines. Cloud
computing is enabled by virtualization. Further research is
needed to better understand virtual machine operation and
establish safeguards to effectively protect this evolving
environment.
Second is authentication, authorization, and accounting.
Current security approaches leverage current best practices.
Research is needed to counter the many threats, including
eavesdropping and tampering, distributed denial of services,
network infrastructure vulnerabilities, and insider threats.
Third, R&D on security applications and tools should focus
on the creation of applications that leverage the distributed
nature of the cloud to provide a new level of security. This
research would result in a more secure environment that is
resistant to both infections of individual hosts and the
current generation of network-based attacks.
Another area is encryption for programs and data
processing. Recent work has produced an encryption system
allowing computers to execute encrypted programs.
Research on distributed denial of service detection and
control is also needed. A DDOS attack is an attempt to make a
computer resource unavailable to its intended users. Currently
there is not a good mechanism for DDOS detection and control.
Finally, research on network technologies is also
important. Current protocols and tools in place today make it
difficult to make networks available dynamically to match the
elasticity in clouds. Adaptive and intelligent networking
research is an important area of study.
It is also critical that we have a security-conscious
workforce. There is a gap that exists between what universities
teach and what industry needs. Universities teach theory and
fundamentals, whereas industries desire practical experience.
In addition, Mr. Chairman, I believe attention should be
given to legal issues surrounding cloud computing--contractual
and service-level agreement issues regarding physical data
protection, incident response, confidentiality, privacy and
security controls, and other matters, which are important
aspects in developing a relationship with a provider.
Mr. Chairman, on behalf of Clemson University, I would
again like to thank you for your time.
[The statement of Mr. Bottum follows:]
Prepared Statement of James R. Bottum
October 6, 2011
Mr. Chairman, I would like to thank you and the Members of the
subcommittee for this opportunity to present testimony before this
committee. I would like to begin by taking a moment to briefly acquaint
you with Clemson University.
Located in Clemson, South Carolina, Clemson University \1\ is a
Nationally-ranked, science and technology-oriented land grant public
research university founded in 1889, known for its emphasis on
collaboration, focus, and a culture that encourages faculty and
students to embrace bold ideas. Clemson's teaching, research, and
outreach are driving economic development and improving quality of life
in South Carolina and beyond. With an enrollment of 19,500, Clemson is
a high-energy, student-centered community dedicated to intellectual
leadership, innovation, service, and a determination to excel.
---------------------------------------------------------------------------
\1\ Clemson University.
---------------------------------------------------------------------------
Regarding my own background, I have been the vice provost and chief
information officer at Clemson University since July 2006. During my
tenure, Clemson has transformed its network, storage, and computational
infrastructure, including the data center, into a state-of-the-art set
of services benefitting research, education, and public service. We
have been recognized for transformative work in publications such as
Network World, Computer World, and Storage Magazine. Before coming to
Clemson, I was the first chief information officer at Purdue University
beginning in 2001 where I forged a new model for partnering with
research (recognized in a publication by the EDUCAUSE Center for
Applied Research, July 2005). Prior to that, I was the executive
director at the National Science Foundation's National Center for
Supercomputing Applications at the University of Illinois at Urbana-
Champaign. I currently or previously have served on a number of
National committees including the National Science Foundation's
Advisory Committee on Cyberinfrastructure and the Internet2 Board of
Trustees.
cloud definition
Mr. Chairman, many definitions exist to explain what ``the cloud''
actually represents. For purposes of my comments today, a good working
definition should reflect what I believe to be the distinctive
characteristic that defines cloud computing, namely the elastic, on-
demand virtual delivery over the internet of shared services, including
infrastructure and software. By allowing users to share access to
software applications, computational power, networks, and data storage,
cloud computing enables computing infrastructure to be right-sized
while balancing user requirements with the information technology
services actually rendered. Recognizing this shared component is
fundamental to understanding the dynamic effects that are derived from
the cloud.
Also inherent in the cloud model is its flexibility. Multiple
implementation regimes--private, community, public, and hybrid--permit
organizations to select deployment schemes that best meet their needs
and missions. Clouds are not one-size-fits-all. As defined in the draft
National Institute of Standards and Technology Definition of Cloud
Computing.\2\ Private clouds are environments where ``the cloud
infrastructure is operated solely for an organization.'' Private clouds
host and on-demand deliver resources, under the control of the
organization, generally within a firewall. Community clouds are where
``the cloud infrastructure is shared by several organizations and
supports a specific community that has shared concerns (e.g., mission,
security, requirements, policy, and compliance considerations).'' This
shared infrastructure enables the community to share in the cost, yet
also offers a common set of security and privacy policies and
procedures. In Public clouds ``the cloud infrastructure is made
available to the general public or a large industry group and is owned
by an organization selling cloud services.'' Public clouds may be free
or pay-per-use and provide resources that are dynamically provisioned
on a self-service basis. Hybrid clouds are environments where ``The
cloud infrastructure is a composition of two or more clouds (private,
community, or public) that remain unique entities but are bound
together by standardized or proprietary technology that enables data
and application portability.''
---------------------------------------------------------------------------
\2\ Mell, Peter and Timothy Grance. National Institute of Standards
and Technology. ``The NIST Definition of Cloud Computing (Draft).''
National Institute of Standards and Technology Special Publication 800-
145. January 2011.
---------------------------------------------------------------------------
cloud evolution
Cloud computing may be characterized as evolutionary over time.
Cloud computing should not be viewed as revolutionary, since some of
the earliest concepts regarding computer time-sharing and utility
computing came out as early as the 1960s, but did not take hold in our
society until decades later. Past models of computing focused on
utilizing supercomputers, mainframes, and storage devices primarily
owned and operated by a single organization. As the internet and
broadband capabilities expanded, opportunities arose to connect, share,
and leverage these resources by multiple organizations with a common
purpose. Referred to as grids, or grid computing, this model provided
multiple users and various sites access to a shared heterogeneous
computational infrastructure utilized to solve computational problems.
During the 2000s, the cloud concept further evolved as major companies
such as IBM, Google, and Amazon as well as numerous universities and
research organizations began to develop and grow environments.
south carolina cloud example
At Clemson University, our own cloud initiative has coalesced
around what we refer to as the South Carolina Cloud \3\ or ``SC
Cloud.'' SC Cloud represents a collaboration of educational
institutions, IT professionals, commercial entities, and others who
drive cutting-edge research in the areas of computing and communication
infrastructure, data storage and visualization, virtual collaboration,
and education workforce training. In pursuing their research,
participants access a cluster of 61,700 PCs as well as other High
Performance Computing resources and networks to virtually explore new
concepts in a host of critical computing research fields, including:
Data modeling, the hyper-growth in connected devices, surge in real-
time data streams, on-line and mobile commerce, business use of
service-oriented architecture, virtualization, and Web 2.0
applications.
---------------------------------------------------------------------------
\3\ South Carolina Cloud.
---------------------------------------------------------------------------
The SC Cloud initially began as a consolidation effort of Clemson's
on-campus distributed computing resources to improve computing
efficiencies and advance capabilities in research and education. One of
the unanticipated results of this effort was the partnerships that
developed with other South Carolina universities. SC Cloud partners
share a common set of computing and IT services, including networking,
high performance computing, server administration, data storage,
instructional and classroom technology support, monitoring, and
security and privacy. Likewise, higher education also share a common
set of issues and challenges related to these services, including the
economics of supporting and maintaining a growing set of services
during economically challenging times, ensuring an adequate workforce,
and continually modifying the service offerings to meet ever-changing
demands and expectations. Across South Carolina the value of working
together in a shared resource environment was quickly recognized as an
evolving ``work-in-progress'' model that enables institutions to more
efficiently and effectively address computing and information
technology collectively.
cloud benefits
Our SC Cloud experience resonates and echoes many of the benefits
found in cloud computing across the Nation, regardless of the cloud
deployment model. Costs are reduced by sharing the overhead capacity
required for peak loads. Large numbers of standardized hardware enables
next-day parts replacement contacts in lieu of expensive rapid response
time, on-site maintenance contracts. Advantageous hardware and software
pricing is negotiated. Economies of scale allow investment in redundant
cooling, backup power, and other facility infrastructure.
Virtualization and infrastructure management solutions make it possible
to rapidly deploy or remove resources incrementally based on demand.
Researchers focus on research instead of administering systems.
Reliability is improved by locating away from high-risk areas. Energy
use is reduced by eliminating the need for powering and cooling unused
capacity, and energy costs are reduced by locating where power is
cheaper.
There are numerous examples of both public and private entities
that have realized sizable benefits from the adoption of cloud
computing initiatives. GlaxoSmithKline, a leading pharmaceuticals
company, recently deployed a Microsoft cloud solution through a
Deskless Worker Suite to 15,000 of its employees, reducing IT
operational costs by 30 percent while enhancing productivity and
expanding external collaboration.\4\ The U.S. Air Force saved an
estimated $4 million annually on its Personnel Services Delivery
Transformation (PSDT) system by implementing a cloud solution from
RightNow and customers can now find answers from over 15,000 documents
within 2 minutes, a drastic improvement from previous wait times of 20
minutes.\5\ The Department of Energy estimates it will save $1.5
million over the next 5 years in hardware, software, and other labor
costs from implementing a cloud solution at the Lawrence Berkeley
National Lab for its e-mail accounts and from utilizing Google Sites
and Google Docs for its scientific research teams.*
---------------------------------------------------------------------------
\4\ Microsoft Corporation--Case Studies. 2009.
\5\ Kundra, Vivek, Federal Chief Information Officer. State of
Public Sector Cloud Computing. 2009.
* [sic]
---------------------------------------------------------------------------
Another benefit of cloud computing adoption is a company's ability
to better manage its power resources for its IT infrastructure. By
deploying an IBM cloud-based endpoint management solution, Fiberlink--
an innovator in voice, data, and IP networking solutions--achieved a
25% annual growth rate over the last 5 years and has saved an estimated
$500,000 a year from improved power management alone.\6\ A study
concluded this year by Verdantix and sponsored by AT&T estimates that
cloud computing could enable companies to save $12.3 billion off their
energy bills and results in a carbon emissions savings of 85.7 million
metric tons by 2020.\7\ Another study from Microsoft and Accenture
revealed that moving business applications to the cloud could cut per-
user carbon footprints by 30 percent for large, already efficient
companies and as much as 90 percent for the smaller and less efficient
businesses.\8\ Cloud computing is not only beneficial to the companies
themselves that use the technology, but its benefits can extend to the
environment at large because of its decreased dependency on independent
hardware sites distributed across a company.
---------------------------------------------------------------------------
\6\ IBM Corporation--Success Stories. 2011.
\7\ Verdantix Research. ``Verdantix Cloud Computing Report For
Carbon Disclosure Project Forecasts $12.3 Billion Financial Savings For
US Firms.'' 2011.
\8\ Accenture Corporation. ``Microsoft, Accenture and WSP
Environment & Energy Study Shows Significant Energy and Carbon
Emissions Reduction Potential from Cloud Computing.'' 2010 .
---------------------------------------------------------------------------
Our experience with SC Cloud has been that it is a collaborative
mechanism for research, as well as the high-quality, innovative R&D it
is delivering to advance our understanding about virtual environments
in ways that are beneficial to both the public and private sectors. It
is this type of environment that is instructive for framing some of the
key considerations in cloud migration. I would like to share some of
that experience with the committee today, particularly in the areas of
security, scalability, and identity management.
security--clemson university example
Concerns over data theft or manipulation and vulnerabilities to
critical applications are real when contemplating the network security
architecture of the cloud platform. Clemson's Information Security and
Privacy organization mission is to protect the confidentiality,
integrity, and availability of information and informational resources.
The goal is to apply policies, procedures, and controls that are
seamless, transparent, and non-impeding to the organization. Controls
match the risks that exist and ensure the protection of data, provide
redundancy, and include the ability to monitor Clemson's environment.
Security and privacy at Clemson are a shared responsibility, meaning
efforts have been made to educate and raise awareness among faculty,
staff, students, alumni, etc. so that security and privacy become a
natural part of the culture.
The security challenges that Clemson faces are typical of other
higher education institutions and similar to those mentioned in Cloud
Security Alliance's Top Threats to Cloud Computing''.\9\ CSA is a
``member-driven organization chartered with promoting the use of best
practices for providing security assurance within cloud computing.''
CSA's research shows that the top security threats include such areas
as insecure interfaces, malicious insiders, shared technology issues,
account or service hijacking, and unknown risk profiles. We have
implemented a series of policies, best practices, and controls that
provide for increased protection, but know that nothing is 100%
``bullet-proof.'' Staying ahead of the curve of threats and
vulnerabilities is a continual challenge, which Clemson addresses
through a variety of best practices that should be part of any
organization's security strategy.
---------------------------------------------------------------------------
\9\ Cloud Security Alliance. ``Top Threats to Cloud Computing
V1.0.'' March 2010.
---------------------------------------------------------------------------
First among these best practices are human resource procedures. A
criminal background and E-verify check is conducted on all university
personnel prior to their hire and employees are bound by
confidentiality in their work. In addition, establishing a series of
policies and procedures provides a foundation by which Clemson's
security strategy has been developed and lays the framework under which
security operations function. Included topics among the policies are
Acceptable Use, Userid and Password, Network Security, Server
Administration, and Data Center access. Regarding security clearances,
employees needing access either physically or virtually, must be
requested and authorized by supervising personnel based on the
employee's job function requirement. Restricted or secure areas are
protected by monitored and recorded video surveillance and key-card
access. Additionally, the main data center facility has staff on-site
24/7/365. Technical controls are put in place based on the evaluated
risk, a variety and matrix of controls would be deployed that might
include physical or logical network segmentation, Firewall and Access
Control List use, increased and elevated levels of monitoring,
separated Virtual Private Network use, limited availability of access,
and more stringent levels of credential use.
scalability--sc cloud and health sciences south carolina examples
For most organizations, economics is the force multiplier driving
them into cloud computing to realize enterprise efficiencies both in
terms of IT spending and asset utilization. Clemson has been in the
``cloud business'' for over 30 years provisioning Medicaid applications
services to the State and citizens of South Carolina. As previously
mentioned, the SC Cloud evolved into a State-wide consortium of
institutions who either could not afford to address the infrastructure
needs on their own or did not have the expertise to deploy in-house
resources. What once started as a Clemson private cloud need, evolved
into a community cloud where the volume of computing and cloud services
increased, but yet did not result in any service degradation at
Clemson. These institutions realized the economic benefit of fully
participating in the SC Cloud, especially in the context of high-
performance computing, as it enables them access to a set of resources
that are flexible, scalable, and reliable to meet current and future
needs. Institutions participating in the SC Cloud include both public
and private universities, including technical colleges and Historically
Black Colleges and Universities,\10\ or HBCUs.
---------------------------------------------------------------------------
\10\ United States Department of Education--Historically Black
Colleges and Universities
---------------------------------------------------------------------------
Likewise, the SC Cloud further evolved and scaled to provide
flexibility for the Health Sciences South Carolina referred to as
HSSC.\11\ HSSC is composed of six of South Carolina's largest health
systems and the State's largest research-intensive universities. This
State-wide biomedical research collaborative has a vision of
transforming the State's public health and economic well-being through
research as well as education and training of the health-care
workforce. Given Clemson's security strategies previously described as
well as our experience being the primary provider of operational
support to South Carolina's Department of Health and Human Services for
Medicaid transactional processing and eligibility determination, HSSC
determined that the SC Cloud would be a natural fit not only for
infrastructure, platform, and software cloud services, but also for
security as a service. Clemson essentially serves as the Information
Security Office for HSSC by providing the same suite of services
afforded to Clemson, but also applying the same confidentiality,
integrity, and availability philosophies, strategies, controls,
policies, and procedures within a HSSC context. This environment shares
much of the infrastructure utilized by Clemson, yet is segmented in
such a way so as to provide a hybrid cloud that addresses both
Clemson's and HSSC's needs.
---------------------------------------------------------------------------
\11\ Health Sciences South Carolina.
---------------------------------------------------------------------------
Building upon the previously-mentioned security best practices,
Clemson's experiences with scalability has demonstrated four additional
areas of consideration when forging a cloud computing security
strategy. First among these is ensuring a trust relationship is
established between client and provider. Current cloud models are
widely used because they provide economies of scale. They also,
however, outsource data and resource management to third parties.
Clients must rely on the ability of the provider to assure privacy,
accuracy, and availability of information. Developing a trust
relationship, as in the case of HSSC with Clemson, is an important
consideration in ensuring the safety of data. Clemson's experience with
Medicaid data as well as the policies, procedures, and controls that
are put in place enable an increased level of trust. Continual
interaction and engagement has resulted in Clemson being at the table
when HSSC is in the early stages of application development and the
subsequent change management. This has resulted in security and privacy
being an integrated, proactive part of HSSC's planning and operations.
Clemson University's relationship with HSSC members has been
strengthened with their deployment of previous investments in
authentication research and development. Clemson University is a
participating member of Internet2's InCommon federated identity
management supporting Shibboleth authentication. HSSC systems has
utilized Shibboleth authentication to allow for multiple trusted
participating members to leverage their own identity management vetting
process and procedures for access to HSSC systems. This is a great
example of how R&D has produced a viable, productive application and
methodology to achieve greater efficiencies and ease of use without
compromising the security of the system.
Second, the level of cloud integration should be considered.
Depending upon an organization's mission and requirements, an
organization may only take advantage of cloud infrastructure services.
Some may pursue software as a service. Yet others may outsource the
entire suite of cloud services, including security as a service. In the
case of HSSC, the SC Cloud provides infrastructure, platform, and
security. In other words, one size does not fit all and a cloud
provider should be flexible.
Third, natural disasters such as Hurricane Katrina, the recent
earthquake in Japan, and the Midwest floods show the importance of
disaster recovery and business continuity. Documenting a plan and
implementing redundancy technologies are obvious components of this
planning. Conducting test failovers and actual physical disaster drills
on a periodic basis should also be included in any DR/BC strategy. Many
lessons are learned when physically conducting a disaster exercise that
enable an organization to be better prepared.
Fourth, one of the reasons HSSC chose Clemson is because of its
Medicaid provisioning experience with medical data, compliance, and
audit response. Clemson has a proven track record of being able to
address internal and external audit requests and quickly address any
findings. A cloud service provider should be able to address their
experience and capabilities in dealing with Federal compliance and
audit needs.
identity and access management
Considering the diverse set of users that the SC Cloud has and the
numerous organizations that connect into the environment, it is
important to properly ensure identity and access management (IaAM).
Identity and access management concerns the need to permit access to
enterprise resources only to authenticated users, with access to only
the data they have permission to view or change. Without appropriate
procedures in place to verify access, concerns over identity theft and
the insider threat can arise.
Authentication is performed when a computing session starts. In
existing systems, a user is authenticated in one of three ways:
Knowledge, which is something the user knows such as a password;
possession, which is something the user has such as a smart card; or
identity, which refers to biometrical aspects, such as a fingerprint.
Clemson's experience has been that identity and access each can be
problematic. Passwords can be forgotten, sent over the network in
clear-text, so that they are readable in transit or revealed
inadvertently. Simple passwords are easy to guess. Complex passwords
are easily forgotten, or need to be written down. Taking IaAM issues a
step further, smart cards, dongles, or other authentication tokens can
be stolen. Voiceprints may have false negatives if the user has a cold.
People are hesitant to use retina scans, since they seem invasive.
Biometrics can also be spoofed. Clemson limits these challenges by
requiring complex passwords, providing training to faculty, staff, and
students, and using a single-sign-on service that forces password
encryption in transit over the network.
On a local machine, authentication is straightforward. If
authentication uses knowledge, for example a password, the user is
prompted directly for the information. If possession is used, the token
(ex. smart card) can be interfaced directly to the computer. Some
authentication systems give the user a device that displays a code
value to enter into the system. For biometrics, a physical device has
to interact with both the user and the computer system. Two-factor
authentication uses more than one authentication technique. This helps
minimize the damage caused by key-loggers and related tools.
All these approaches assume the device used to access the internet
is trustworthy. If the local hardware or software is not trustworthy
(for example compromised by malicious software) this will compromise
both knowledge and biometric authentication.
Access control is at least as challenging as authentication. When
all data and users were locally created and managed, it was relatively
easy to provide controlled access. However, in the cloud, it is more
difficult to provide controlled access. It is possible for there to be
different levels of security for systems and different levels of
assurances for users. The basic infrastructure security level within a
public cloud should match the level of the highest security need, not
be a mixed bag of approaches. Understanding the access control security
practices as well as the results of the provider's risk assessment
efforts are essential considerations. As discussed later in my
testimony, further study is needed in the area of identity and access
management technologies and policy.
considerations
Mr. Chairman, the power of cloud computing offers tremendous
advantages to both the commercial and public sectors. For our
Government agencies in particular, cloud migration represents an
achievable strategy for deriving the tangible cost savings that the
current economic and fiscal environment demand. Furthermore, it enables
both the smart, streamlined organizational construct that Government
employees need to better perform their mission, and the more efficient
services delivery model that taxpayers deserve. And, while I have
enumerated some of the challenges that exist, it is my view that the
benefits of cloud far outweigh the risks, and that a thoughtful
strategy for prudently broadening adoption of cloud services can
facilitate a smooth transition to this dynamic platform. Many of the
security-oriented policies, procedures, controls, and best practices
previously mentioned are key elements of any security strategy.
Additional components that such a strategy might consider include
current areas of research and development, education and workforce
priorities, and economic implications.
areas of research and development
Many areas of research and development exist in the cyber-security
field. It is my opinion as well as the opinion of other researchers in
the field that Cybersecurity R&D is best conducted in an operational
environment as opposed to a simulated environment. The SC Cloud was set
up in an operational environment with this principle in mind. IT staff
provisioners work side-by-side with researchers from academia and
industry across the spectrum. Cybersecurity is critical to all
communities. An exemplary Federal program that includes this program is
the NSF funded Global Environment for Network Innovation or GENI.\12\
Core premises of GENI are that the internet architecture is over 25
years old and in need of strengthening and updating. A second premise
is that network R&D should be conducted on the internet itself and the
GENI approach is to use ``slices''. Analogous to the use of virtual
machines to allow isolated computing on a shared computer, emerging
technologies now allow virtual network slices to be created on shared
network infrastructure to allowed isolated network operation. Network
virtualization not only allows cyber R&D occur on production internet
in protected ways, it also enables isolated and secure enterprise
operations to take place on a shared network.
---------------------------------------------------------------------------
\12\ Global Environment for Network Innovations (GENI).
---------------------------------------------------------------------------
My comments will highlight some research, which in my opinion are
of importance and worthy of investment.
The first area of R&D involves the use of virtual machines (VMs) in
clouds. Cloud computing is enabled by virtualization. This has enabled
servers to migrate from one host to another dynamically for load
balancing as well as made easier dynamic recovery from hardware
failures. Security can be enforced by executing programs on different
virtual machines. Virtual machines, however, are subject to various
vulnerabilities. Researchers at Clemson have shown how power and timing
data can be used to extract information, including cryptographic keys,
from running systems. Further research is needed to establish what
hardware safeguards are required to effectively protect virtual machine
environments.
The second area of R&D is authentication, authorization, and
accounting. Current security approaches leverage current best practices
for authentication, authorization, and accounting relying on Public Key
Infrastructure (PKI) and a certificate authority (CA) hierarchy to
establish a chain of trust. Traditional approaches are designed to
secure monolithic computing entities, but the distributed nature of the
cloud could be leveraged to provide additional security.\13\ As cloud
computing leverages distributed resources at different sites and
potentially of different ownership--for example, an enterprise might
dynamically purchase computing resources from multiple cloud providers
for resilience, load balancing, and cost optimization, the cloud user
needs ways to identify itself in consistent, unified, secure, and
portable means to all resources.
---------------------------------------------------------------------------
\13\ R.R. Brooks, ``Mobile code paradigms and security issues,''
IEEE Internet Computing, vol. 8, no. 3, pp. 54-59, May/June 2004. R.R.
Brooks, Disruptive Security Technologies with Mobile Code and Peer-to-
Peer Networks, CRC Press, Boca Raton, FL, 2005.
---------------------------------------------------------------------------
R&D on security applications and tools is another area of research
that focuses on the creation of applications that leverage the
distributed nature of the cloud to provide a new level of security that
neutralizes security vulnerabilities and the various classes of
attacks. This research would result in a cloud environment that is
resistant to both infections of individual hosts and the current
generation of network-based attacks.
Another R&D area is encryption for programs and data for
processing. Recent work \14\ has produced a true homomorphic encryption
system that allows computers to execute encrypted programs. In theory
this should be free of side-channels, but the newness of this approach
means that vulnerabilities may still be found.
---------------------------------------------------------------------------
\14\ C. Gentry, A Fully Homomorphic Encryption Scheme, Ph.D.
Dissertation, Dept. of Computer Science, Stanford University, 2009. T.
Rabin (ed.) Advances in Cryptology--Crypto 2010. LNCS vol. 6223,
Springer Verlag, Berlin 2010.
---------------------------------------------------------------------------
Research on Distributed Denial of Service (DDoS) detection and
control is also needed. A Distributed Denial of Service attack is an
attempt to make a computer resource unavailable to its intended users.
A DDoS attack can shut down cloud service site or constantly affect
cloud performance, thus increasing the costs. Currently there is not a
good mechanism for DDoS detection and control. It is not possible to
detect the source of the DDoS or control the traffic. DDoS is currently
an intensive area of research. For example, the National Science
Foundation's GENI project funds researchers at Clemson to leverage
OpenFlow, a software-defined networking technique, to flexibly analyze
network traffic for DDoS threats and control different categorized
traffic to ameliorate detected threats.\15\ Some suggestions have been
made for ways to create DDoS-resilient clouds.\16\
---------------------------------------------------------------------------
\15\ Brooks, Richard and Kuang-Ching, Wang. EAGER-GENI Experiments
on Network Security and Traffic Analysis. National Science Foundation
Award No. 1049765.
\16\ Dingankar, C. (MS) ``Enterprise Security Analysis Including
Denial of Service Countermeasures,'' ECE Dept. Clemson University
(August 2007). C. Dingankar, S. Karandikar, C. Griffin, and R.R.
Brooks, ``On Bandwidth Limited Sum of Games Problems,'' IEEE
Transactions on Systems, Man and Cybernetics, Part A: Systems and
Humans, 41(2) 341-349, March 2011.
---------------------------------------------------------------------------
Finally, research on network technologies is also important.
Current protocols and tools in place today make it difficult to make
networks available dynamically to match the elasticity in clouds.
Networks tend to be static and specialized with data passing through
hundreds of thousands of separate network devices that operate
individually instead of as a unified system. A paradigm shift is needed
to instill more dynamic control plane flexibility to match the growth
of diverse applications and devices utilizing cloud services, including
mobile, across entire networks in a cloud environment.
Such a paradigm shift can be seen today through the implementation
and use of Software Defined Networking (SDN) technology such as
OpenFlow,\17\ which has been developed as the network layer of the GENI
model. SDN moves the control plane from the individual network device
to external controllers that can view and manage a network as a system
instead of a vast network of individually-configured devices.
Additionally, SDN makes it easy for new network protocols to be rapidly
prototyped into production networks.
---------------------------------------------------------------------------
\17\ OpenFlow.
---------------------------------------------------------------------------
In addition, adaptive and intelligent networking that does not rely
only on the end-host or individuals for correct protocol application is
an important area of study. One cannot rely on all providers having
firewalls, consistent security standards, intrusion detection, etc.
Distributed tools are needed to enable automated security through
improved network monitoring to analyze traffic patterns and detect/
isolate vulnerabilities as well as securing internet traffic in
distributed and seamless ways.
education/workforce priorities
Mr. Chairman, in addition to R&D, it is also critical that we have
a security-conscious workforce. There is a gap that exists between what
universities teach and industry needs. Universities teach theories and
fundamentals whereas industries desire practical experience from
university graduates. This is difficult to incorporate into the
curriculum. Programs are needed to facilitate bridging this gap and
partnerships between universities and 2-year technical and community
colleges should be encouraged. In addition programs that encourage
students to major in science, technology, engineering, and mathematics
(STEM), including an emphasis on cyber-security, are needed.
NSF GENI is an example of program that is filling this gap by
creating an environment linking industry with university research thus
providing experiences for students to receive training and education on
core technologies that are applicable in the workforce. In addition,
GENI also extends these opportunities to multiple disciplines ranging
from computer software, computer system, networking, to hardware
engineering thus giving a student a broader experience of conducting
research and having regular interaction on a large scale with other
fields of study. Federal facilitation of similar programs in cross-
cutting areas may begin to close this gap over time.
economic implications
There is a growing body of research involving interactions between
information security and economics.\18\ Current market incentives
reward behaviors that do not safeguard the well-being of the public.
This is in direct conflict with the Institute of Electrical and
Electronics Engineers (IEEE) \19\ and Association for Computing
Machinery (ACM) codes of ethics.\20\
---------------------------------------------------------------------------
\18\ Anderson, R. and T. Moore, 2008: Information security
economics--and beyond. Lecture Notes in Artificial Intelligence, 5076,
49.
\19\ Institute of Electrical and Electronics Engineers Code of
Ethics
\20\ Association for Computing Machinery Code of Ethics and
Professional Conduct
---------------------------------------------------------------------------
Hardware and software markets have network externalities: The value
of an investment depends in large part on whether or not other parties
make the same purchase decision.\21\ These markets are ``tippy,'' i.e.
miniscule differences in quality or perception result in major
differences in profitability. In our industry, network externalities
often result in markets where one product dominates the market. This
explains the historically dominant market positions of the IBM PC,
Microsoft Windows, and Intel processor architecture.\22\ The need to be
the dominant player induces pressure to be ``first to market'' with new
applications. Arriving early usually tips the market enough to dominate
it. In this ``winner take all''\23\ context, actions that improve
product quality and security, but delayed delivery can be fatal to an
enterprise.
---------------------------------------------------------------------------
\21\ Katz, M.L. and C. Shapiro, 1985: Network externalities,
competition, and compatibility. The American Economic Review, 75, 424-
440.
\22\ Besen, S.M. and J. Farrell, 1994: Choosing how to compete:
Strategies and tactics in standardization. Journal of Economic
Perspectives, 8, 117-131.
\23\ Dekel, E. and S. Scotchmer, 1999: On the evolution of
attitudes towards risk in winner-take-all games. Journal of Economic
Theory, 87, 125-143.
---------------------------------------------------------------------------
This is exacerbated by software being a ``lemon market''\24\ with
information asymmetry between buyer and seller. The buyer cannot
reliably distinguish between quality goods and shoddy products. Under
these conditions, buyers choose the lower-priced product. Shoddy
products are produced more cheaply, driving quality products from the
market.
---------------------------------------------------------------------------
\24\ Akerlof, G.A., 1970: The market for ``lemons'': quality
uncertainty and the market mechanism. The Quarterly Journal of
Economics, 84, 488-500.
---------------------------------------------------------------------------
These factors encourage the industry to quickly produce large
quantities of poorly analyzed programs. There is little financial
incentive to do otherwise and much to gain. The consequences of poor
software quality for consumers and the economy as a whole are immense.
Dr. David Rice cites NIST studies showing the annual cost of insecure
software to the United States as conservatively $180 billion.\25\ He
also cites a market research survey, which finds 75 percent of
computers connected to the internet have been infected and used to
distribute spam. Computer and network security is likely to remain a
difficult problem for the foreseeable future. Research and development
of secure systems will be costly, but that cost is expected to be much
less than current losses due to on-line system misuse.
---------------------------------------------------------------------------
\25\ Rice, D., 2008: Geekonomics. Addison-Wesley, Upper Saddle
River, NJ, 2nd ed.
---------------------------------------------------------------------------
other considerations
In addition, Mr. Chairman, there is one other priority that I
believe will receive attention as cloud services grow, namely the many
legal issues surrounding cloud computing. Contractual and service-level
agreement issues regarding physical data protection, incident response,
confidentiality, access, availability, privacy, security controls, and
other such critical matters are important aspects in developing a
relationship with a provider. Likewise, intellectual property issues
and export controls, meaning where is the data being stored, should
also be discussed in a cloud computing strategy. It is conceivable that
some cloud service providers could store data outside the United States
for backup or archival purposes. Also, consideration should also be
given to the portability of data and what happens to the data once a
provider contract is terminated. Safeguards and assurances are
important to ensure all data packaged for migration to a new provider
and that all data is cleaned and removed from any provider asset.
Finally, considering the level of hardware manufacturing that occurs
overseas, assurances that personal computers, tablets, etc. do not
contain viruses or other security compromising elements is needed.
Mr. Chairman, on behalf of Clemson University I would again like to
thank you for the opportunity to testify before the subcommittee and I
look forward to your questions.
Mr. Lungren. Thank you much, Mr. Bottum. I was just
thinking that cloud computing is the only thing I have not
heard being argued for the breakup of the Big East or ACC, and
I suspect that maybe we will be hearing about that----
Mr. Duncan. Will the gentleman yield?
Mr. Lungren. Sure.
Mr. Duncan. Go Tigers against Boston College----
[Laughter.]
Mr. Lungren. Well, I have got a neighbor who is a freshman
at Clemson, so I will say okay.
Mr. Curran.
STATEMENT OF JOHN CURRAN, CHIEF EXECUTIVE OFFICER, AMERICAN
REGISTRY OF INTERNET NUMBERS
Mr. Curran. Thank you, Chairman Lungren, Ranking Member
Clarke, Members of the subcommittee, for having me here today.
You have my written testimony so I will keep my verbal comment
brief.
I am going to focus on areas related to using the cloud
over the public internet, because that is truly what is new in
what we are discussing. Dr. McClure, earlier today, indicated
that the use of public clouds poses new areas of risk, and I
would like to highlight four of those that this committee
should consider when looking at this issue.
First is, the relationship of public clouds to other
initiatives within the Federal Government for cybersecurity
needs to be carefully considered, because public clouds are
using vendors outside the Federal Government, yet the Federal
Government has several Government-wide security initiatives.
These include HSPD-12 for validation and authorization; this
includes the Trusted Internet Connections program.
When you make use of a public cloud and a public vendor
they may not be familiar with how to actually use those
initiatives, which are Government-wide cybersecurity
initiatives. So the documentation and the approach to vendors
so that they have everything they need when they design their
public cloud to make use of Government-wide cybersecurity
initiatives is essential. Otherwise, our public clouds won't be
participating in our Government-wide initiatives. This is very
important.
Second is the issue of the physical location of the actual
data and systems. The FISMA framework and the FISMA security
control profile always had an assumption within it of Federal
control of the facilities or systems. It is true about 10
percent of our Federal inventory is outsourced to contractors,
but even then, it is outsourced in a way that puts it directly
under agency control to the vast majority of cases.
When we make use of public clouds we suddenly have the idea
of using a FISMA profile that is 10 years old to secure public
clouds that may actually be worldwide in nature. The problem,
of course, is that the questions to be asked--where is the
data, where is the systems--simply don't exist in the original
FISMA profile.
Now, the proposed FedRAMP security profile does have
enhancements, and one of the enhancements it includes is
talking about the personnel that are making use of managing
this data. In the current public drafts it does not include,
however, controls for where is the data and the systems
themselves? So we know, in many cases, that the systems are
managed by U.S. citizens, but we don't know necessarily that
they are located within the United States.
A given agency can implement SLAs to cover that if they
know to do so. What might be a better approach is making that
inherently part of the profile, so as GSA accredits
organizations they say where there systems are, so a Federal
agency CIO has the ability to say: Is that acceptable to me or
not?
The third matter is on migration, and I guess this is more
important. The FISMA profile is very good about talking about
recovering of systems; it has a whole contingency planning
section which handles the failure of a given server or data
center. That was perfect for when we were talking about Federal
agencies.
But the recovery now that is provided by the FISMA profile
now works within the cloud. It is whether a cloud provider
provides fail over one of their data centers to another one of
their data centers.
The problem is, we now need contingency planning at one
level higher up. In fact, you need to worry about the case
where the cloud provider themselves is no longer able to
provide service securely and you need to move not to another
one of their data centers but to an entirely different cloud
provider. You might need to do that on very rapid notice to
accommodate a cloud provider who is compromised in an
irrecoverable manner.
So the migration is not a question just of cost or being
able--agencies being able to get their own data back. It is
actually a security control. It is an inherent function that
needs to be provided so that if a cloud provider is compromised
the ability to migrate isn't a question that we are all asking;
it is inherent and it is known to be able to quickly move up in
a short number of days or hours and move to another provider.
Finally, the internet itself: The internet itself is not
static. It is changing rapidly, and there are several security
protocols, such as DNSSEC, to secure the Domain Name System,
and I.P. version 6, the new internet protocol that is coming
out, that need to be considered. We need to make sure these are
part of the profile for FedRAMP so we don't build on the
internet while the internet is changing out from under us.
I would like to thank the committee for having me. I look
forward to your questions.
[The statement of Mr. Curran follows:]
Prepared Statement of John Curran
October 6, 2011
i. introduction
Good morning Chairman Lungren, Ranking Member Clarke, Ranking
Member Thompson, and Members of the committee, and thank you for
inviting me to testify before the Cybersecurity, Infrastructure
Protection, and Security Technologies Subcommittee.
I am the president and chief executive officer of the American
Registry for Internet Numbers, Ltd. (``ARIN''), which issues Internet
Protocol (IP) number resources for the United States, Canada, and
Caribbean, but I am speaking here today in my personal capacity based
on a long history of building and securing FISMA-compliant Federal
Information Technology (IT) systems.
I have first-hand knowledge of these matters from my experience in
the internet industry since 1990, including serving as the chief
technology officer for several Government contractors and Internet
Service Providers (ISPs) including BBN, GTE Internetworking, and XO
Communications, as well as internet standards work in the Internet
Engineering Task Force (IETF). Most recently, I served for 5 years as
executive vice president and chief technology officer for ServerVault,
providing secure managed IT services for sensitive Federal Government
applications. My duties included direct responsibility for securing and
preparing the certification of FISMA Moderate impact level Federal
information systems over shared internet-based infrastructure. I have
prepared my remarks today out of a desire to see the advancement of
responsible Cloud-based computing for the Federal Government.
I would like to start by offering congratulations to the GSA for
the development of its Federal Risk and Authorization Management
Program (FedRAMP) program, as well as the recent Infrastructure as a
Service (IaaS) Blanket Purchase Agreement (BPA) awards. By developing
this program in cooperation with the Federal CIO council, the GSA has
enabled agencies to leverage cloud-based storage, virtual machines, and
web hosting services in a manner that should improve the cost and
timeliness of Federal IT system deployments.
ii. managing emerging risks from cloud computing
As a result of my experiences deploying Federal IT systems over the
public internet, I was asked to present at cloud interoperability
workshop in 2009, and to identify the most critical challenges that
Federal CIO's faced in making use of cloud computing under the existing
FISMA security framework. Back then, the major difficulties that I
identified were:
Agency pressure for deployment of timely, cost-effective IT
systems;
Administration expectations for leveraging new IT
technologies;
Compliance with IT policy mandates (Federal and agency-
specific);
Lack of common IT infrastructure services between systems &
Potential vendor lock-in with any sizable deployment;
Preparation of extensive FISMA control documentation for
each system.
It is remarkable to see the progress that has occurred since that
time. As a result of the FedRAMP program (with its common security
control baseline), agencies now have a clear roadmap that should
address many of these challenges in making use of cloud computing for
Federal IT applications.
I must note, however, that cloud computing does not eliminate all
of the challenges, and in particular, cloud computing may actually
heighten the difficulties that Federal CIO's face in some areas if not
carefully managed. The areas that are most likely to pose increased
risks as a result of the introduction of cloud computing are:
1. Interaction of cloud computing services with Federal
cybersecurity initiatives;
2. Physical location of cloud computing facilities and data;
3. Migration between vendors of cloud computing services;
4. Evolution of cloud computing services with internet
technologies.
None of these risks precludes the use of cloud computing services
by the Federal Government, but each does pose new challenges for
Federal CIO's to consider and may warrant consideration by the Federal
CIO Council and its partners to determine if additional standards or
coordination activities would help minimize these risks. I will outline
each of these risk areas with recommendations for further
consideration.
iii. interaction of cloud computing services with federal cybersecurity
initiatives
There are several Government-wide IT security initiatives that
require consideration with respect to cloud computing because of their
service nature: Specifically, there is the distributed issuance and
recognition of user authentication credentials via the HSPD-12
initiative, as well as the provision of secure and monitored internet
connectivity via the Trusted Internet Connections (TIC) initiative.
These programs provide certain security-related services to Federal IT
environments which result in increasing cybersecurity protection on a
Government-wide basis as more agencies make use of the services.
While specified in the FedRAMP security profile for Moderate risk
environments, the actual mechanism and ability to participate in these
Government-wide cybersecurity initiatives by private cloud computing
vendors remains unclear, and any deployment of Federal IT systems via
cloud computing services that do not leverage these common capabilities
dilutes the value of these initiatives in supporting the overall
cybersecurity stance of the Federal Government.
The goal must be to have unequivocal documentation for cloud
computing companies on how to appropriately secure their offerings,
including how to make use of Government-wide cybersecurity initiatives,
and thus encourage significant industry-wide vendor participation in
offering FedRAMP cloud services. The resulting competition will both
drive down costs and improve service quality for all FedRAMP
participants.
iv. physical location of cloud computing facilities and data
One of the more unusual consequences that results from the use of
the cloud computing is the potential loss of the ability to know at any
given time the specific physical location for the systems and data
which support a given Federal IT system. While it may be possible to
know the set of data centers which support the service (and the FISMA-
based FedRAMP security control profile does specify certain physical
controls at such facilities for facility access, power redundancy,
etc.), the question of actual physical location of the Federal IT
system is highlighted when the cloud service provider has facilities
which are outside of the United States.
As a practical matter, there may not be a concern with incident
services being provided for out of non-U.S. locations, and it may be
desirable in some circumstances with Federal applications that must be
accessed globally. However, the present FedRAMP profile does not
directly address the question of location and it is not assured that
use of facilities and storage of data outside the United States is
universally desirable, particularly if the use of cloud computer for
Federal IT applications is undertaken on a large scale.
The FedRAMP program should include controls that address the
physical location of cloud computing facilities and data storage used
by the application, and allow (as is done with the corresponding
personnel controls) for the consideration of exceptions once fully
documented and reviewed.
v. migration between vendors of cloud computing services
The ability to extract agency data in standard formats from cloud
computing services (whether that be application data such as mail
messages and mailing lists, or system data such as the virtual server,
storage, and network configurations) is essential to be able to migrate
between cloud vendors. Lack of this capability means vendor lock,
eroding the financial benefits of cloud computing and preventing timely
response if a vendor's security is irrevocably compromised.
There are on-going efforts in the area of standards for cloud
computing data, and this work should continue and be prioritized by the
agencies supporting the FedRAMP program. Unlike an internal agency
information system, cloud solutions are inherently subject to change by
the cloud service provider, and this creates a new requirement
(specifically, the ability to quickly and reliably migrate to another
service provider) where it previously was not needed for agency
systems. FedRAMP must facilitate migration capabilities to protect
against any cloud computer vendors that fail to continuously deliver
the necessary quality or security in their offerings.
The FedRAMP security control profile includes standard FISMA
contingency planning and recovery security controls, but these
fundamentally only address recovery within a given service provider
cloud. Specific mechanisms should be put in place to insure that
Federal agencies can extract their data and configuration in generally
accepted formats and that these mechanisms suffice for service
migration to other cloud computing vendors.
vi. evolution of cloud computing services with internet technologies
The internet is constantly evolving with the introduction of new
standards and technology, and in making use of the internet as a
platform for cloud computing, FedRAMP must be equally prepared as these
changes occur. This is particularly true when it comes to internet
technology improvements in the area of cybersecurity.
In many cases, the Federal Government has taken an active interest
in the technologies and standards that could improve the overall
security of the internet, and this includes DNSSEC initiative in
securing the Domain Name System (DNS), the next version of the
underlying network protocol for the internet--Internet Protocol version
6 (IPv6) and on-going work in internet routing security. These
technologies are now being deployed in the public internet, and are
also covered by specific directives in the FISMA security control
baseline and/or guidance from OMB.
These new standards are quite important in protecting the global
internet from cybercrime, in that they insure that internet users reach
the actual website that they intended to, and that their communication
is protected in the process. When it comes to agency use of cloud
computing services, these protections are equally important, since
these services are reached over the public internet.
It is crucial that the FedRAMP program clearly and unambiguously
incorporates DNSSEC and IPv6 into the FedRAMP baseline, and that on-
going developments in internet-wide security technologies are promptly
incorporated as they reach maturity.
Furthermore, the on-going need to adopt and maintain state-of-the-
art security technologies and practices for cloud computing services
does not appear to be given sufficient priority in the FedRAMP
approach. While traditional Federal IT systems have been built and
certified one at a time in predominantly closed environments, the rapid
pace of evolution of internet threats requires equally dynamic and
responsive security responses. Vendors should be given the flexibility
to propose additional or alternative security mechanisms, as there are
security lessons learned from running large-scale internet services
that are not readily available to the Federal IT community, and the
benefits of such experience should not be lost in the process of
structuring cloud services into the FISMA framework.
vii. conclusion
The FedRAMP program is a remarkable achievement; by providing
agencies with ready access to additional computing resources that have
already undergone a joint authorization process, the program offers the
potential to significantly improve cost and timeliness of Federal IT
deployments.
While not detracting from the importance of this achievement, the
use of public and shared cloud computing services does introduce new
areas of risk to be considered, and this is particularly true with
respect to the interaction of cloud computing services with Federal
cybersecurity initiatives, the geographic location of Federal data, the
migration between vendors of cloud computing services, and the
evolution of cloud computing services with the internet.
The risks should not preclude use of cloud computing services by
the Federal Government, but the model should be closely examined, and
appropriate efforts inserted into the FedRAMP program so that it can
deliver its full benefits in an efficient and secure manner.
Mr. Chairman, Ranking Member Thompson and Members of the
subcommittee, this concludes my written statement.
Thank you again for this opportunity to speak before you today on
this important topic, and I would be happy to answer your questions.
Mr. Lungren. I thank you, Mr. Curran.
I thank all of you for your testimony, and I will yield
myself 5 minutes for first questions.
Mr. Sheaffer, one of the things that struck me as you spoke
was the idea that in the past, with the internet and so forth,
we didn't build in security at the outset and we have had to
play catch-up. Mr. Curran has just outlined a number of things
that deal with building security into our advances in computer
technology, including the cloud. Could you comment on the
comments that he made?
Mr. Sheaffer. Certainly, sir. I agree that we are in a
position where we are using a technology and infrastructure
that was not originally intended to be with the security issues
in mind, and I agree that there are a number of initiatives
underway to address a number of those vulnerabilities and
issues.
I think there has been a--there is good examples that exist
in--within our intelligence community and in the secure side of
Government operations that point the direction that we are able
to build architectures that can secure data and applications
adequately in a private cloud environment. I think some of the
comments were addressed to how are we going to do that in the
public environment, and I would go back, I think, to some
comments in the earlier panel that suggest that until we can do
that we have to be careful about what we put out into the
public domain.
But the interest of the commercial sector is to, as quickly
as possible, get to a point where they can provide those
adequate protections and the innovation that is going on in the
commercial world, I think, will solve those problems in time.
I think in the mean time I would agree, we have to be aware
of what they are, do what we can from a standards perspective
to build in standards and approaches that will guarantee to the
maximum extent possible that those vulnerabilities can--and
risks can be managed. But we will, as a--from a technological
perspective, solve those problems.
Mr. Lungren. Mr. Brown, it appears one of the messages from
this panel is that the dynamism of the I.T. world----
Mr. Brown. Yes.
Mr. Lungren [continuing]. That we make a mistake when we
take a static view of things and that cloud computing is one
evolutionary point in this utilization of advanced information
systems. So therefore, we have got to try and, from our
standpoint, make decisions that reflect that.
At the same time, there is this sort of fundamental issue
or concern that reflected in both constituencies and Members of
Congress that there is something about possessing a system,
there is something about possessing your own information, there
is something about fencing off your information from everything
else, which is perversely at odds with using the internet.
Mr. Brown. Right.
Mr. Lungren. Yet, people seek both the ease of access and
the multiplication of recipients of their information that the
internet offers with a heightened sense of privacy. So I think
one of the great concerns we have to deal with--both legitimate
aspects of it and, let's say, hyped aspects of it--are that as
you surrender your possession of the system and move more to a
cloud system, which, as I get your various definitions,
essentially means you are cooperating with other systems in a
way that your information is not totally under your control,
how do we both overcome the fear that people have a loss of
security because of a loss of possession, but at the same time
assure them that we do have technology fixes so long as we
understand that that requires a sufficiency of information that
the users have and a persistence in the use of what I will just
call generally good cyber hygiene?
Mr. Brown. So, one of the things that we have to understand
is that from an economic standpoint, cloud is coming, okay? The
reason why is that in cloud computing we can do many more
releases, put together more software that is better more
quickly, we can test it in one environment, we can get higher
quality software out of the, you know, out of our building and
into the hands of the consumers quicker.
If we don't, as vendors, embrace cloud we will be out of
business, okay?
Mr. Lungren. That is a pretty strong imperative.
Mr. Brown. Yes. So it is a very strong imperative. If we
don't embrace cloud we will be out of business.
So I think the same goes for governments in the same way,
that if you want to keep up, if you want to move quickly,
embracing the cloud for the same efficiency reasons needs to
occur. Now, anytime we have these types of changes, right, we
have opportunities to become better or become worse, right? We
believe that cloud gives us an opportunity to become more
secure.
Now, the things that need to happen there is you need to
have trust in the providers, like as what we said, but you need
to be able to verify, right? So you need to be able to have
things like FedRAMP that allow you to monitor those providers
to make sure they are not only doing what the contract says,
but actually doing what they say, right?
You need to be able to be cautious as you go in--enter into
these environments to make sure that--you know, in some cases
we are going to see huge expansions of cloud providers and only
a certain portion of them will survive, so you need to have
contingency plans set up to be able to move from one cloud
provider to another.
So it is not a question of if it is going to happen. It is
going to happen; we are going to move there.
So it is a question of how we get enough trust in that
environment that we can effectively move forward. Trust ends up
being transparency; it ends up being, you know, acceptance of
this is what a--this is what a cloud provider is going to do;
and the ability to consistently monitor what they are doing to
ensure that, you know, what they--they are doing what they say
they need to do.
Mr. Lungren. I have got a whole bunch of other questions,
but I am going to yield to Ms. Clarke now for 5 minutes.
Ms. Clarke. Thank you very much, Mr. Chairman.
I want to thank the panelists for lending their expertise
to this discussion today.
My first question is, many potential agency users of the
cloud believe it is not yet secure enough for their needs. From
your perspectives, are they right?
Mr. Bottum. Well--excuse me--I am a provisioner, and so I
say amen to everything Mr. Brown just said, and it is a
question of building up trust. I think with the relationships
we have, you know, that is essentially how we got there, was
through building the trust of the end-user and the community
that we are provisioning for.
The first thing I did 5 years ago when I went to Clemson
was consolidate 43 I.T. departments into one, and that is
essentially building a cloud for 43 people who used to do their
own--departments that used to do their own computing. So over
time you have to, you know, build that trust and that true
performance.
I think, you know, directly answering your question is it
secure enough, we get tested in a number of ways. I think the
end-user has to figure out how they, you know, trust but
verify, I think.
I mentioned that we run the Medicaid system for the State
of South Carolina. We get both planned visits, audits, and we
get unplanned visits and audits. So you have to be ready at all
times.
It is a matter of communication, policies, people working
together. I think, you know, the--to me, you know, the cloud
is--you know, we just call it something different every decade.
It was time-sharing in the 1980s; it was the grid in the 1990s.
We did a project with Notre Dame, the Northwest Indiana
Computation Grid.
But basically it is, you know, that is essentially what it
is, is a matter of people working together and creating a
trusted environment, so----
Mr. Curran. Let me address this a little bit, and I will
pick up on the comments of the earlier panel from DHS CIO
Spires. At the end of the day, the question of whether or not
secure enough is the agency CIO's determination. That is truly
his job.
What we need to do is make sure that the mechanisms we have
put in place give that agency CIO enough information to make
that determination. The FedRAMP program is a start at a profile
of controls that would make public clouds useful to CIOs.
However, right now there are a number of pieces that a CIO
has to fill in on their own. If you want your data within the
United States that is not in the profile; that is your SLA. If
you are worried about migration, that is not in the profile;
that is something you are going to worry about.
So the answer is: Is it suitable today? For an ambitious,
high-energy agency that decides it is going to take this on,
yes, where they fill in those pieces. So the question is
whether or not we can make a FedRAMP program where those
functions are already provided for, already clearly documented.
That doesn't mean all the data, for example, needs to be in
the United States. An agency whose workers are around the globe
doing aid might want data centers that are close to where those
people are for performance reasons. Someone else doing
sensitive work might want to know that the cloud that he is
using has said all of its servers are located in the
continental United States.
It is making sure that information is in the profile and in
the documentation so the agency CIO has the work he has, has
the information he needs to do the job of answering that
question. I think it is possible to use it today; I think it
could be much easier to use with work.
Mr. Brown. One of the other important points here is that
there will be specialized cloud services that are developed for
specialized purposes, okay? So, you know, if there is enough
money available for someone to produce a cloud service that is
ultra-secure, you know, ultra--you know, ultra-secure and
ultra-resilient, right, somebody will produce that cloud
service from--as long as the economic model fits.
You are going to see other economic models that take less
security, and less security less resilience. All of those types
of models are okay as long as they transparently tell you what
their models are and what they can provide.
Ms. Clarke. Let me thank you all for your answers. So many
questions come to mind once you raise that question and then
you get the answers, right? So it is a totally new space. That
is a lot of pressure on a CIO.
Then you start thinking about, well, does this become an
issue for litigation as, you know, we begin to build those
areas of trust, all right? So does that become a whole 'nother
practice within the legal field and an understanding of that
world that we have created?
So, my time is elapsed and I want to just thank you once
again for raising the consciousness here in the Congress of
what we need to do. Thank you.
Mr. Lungren. There are so many questions, but you have been
very good about--let me just ask one general question. When we
look at all the positives of cloud computing, however we want
to define it, and as the new evolutionary point, is it a canard
to suggest, though, that with cloud computing you do create
some more target-rich environments? That is, if I could go
after a larger bit of information or a larger universe of data
points that involve a number of different players it might be
worth my while to put more capital investment and time to go
after it, or is that just----
Mr. Brown. Same idea as Fort Knox, right? So can we protect
the gold, right? That is the question, right, is: Can we have
appropriate safeguards to protect that information?
If you look at what some systems have done, you know, your
data actually isn't stored in one central location; little
pieces of your data are stored all over, in many different
servers all over the, you know, world, therefore they can't be
reconstituted into one piece. So, you know, because the data
just happens to be stored in the cloud it takes advantage of
technology that makes it harder to compromise one data center.
It won't help you. You have to compromise the whole system.
So there are technology advantages to, you know, moving to
the cloud. But you are right about a target, right? As you have
more data in one place it is more of a target, but it is also
one of the things that you can centrally protect.
Mr. Lungren. Well, I want to thank all four of you for
testifying, and the previous panel. This is an issue that we
are just scratching the surface on here. I think there is a lot
of confusion about it, I guess even, I would say, fear, just
because this is a new notion to the larger public, computer--
cloud computing.
I think one of our obligations is not only to help clear up
that confusion as best we can, but understand the reality as
best we can.
I think what you suggest, Mr. Curran, is make sure that all
the moving parts are related, that if we do something on the
Government side where we think we have certain protections that
that is not only communicated with but is operational with
public clouds as we work with them, and that we sort of
anticipate these things instead of doing patchwork approaches
later on.
So I want to thank you. I thank you for your valuable
testimony.
Members of the committee may have some additional questions
for you and, we would ask you if you would please respond to
those in writing upon your receipt. The hearing record for
Members will be held open for 10 days, and the subcommittee
stands adjourned.
[Whereupon, at 12:56 p.m., the subcommittee was adjourned.]
A P P E N D I X
----------
Questions From Honorable William Keating for Richard Spires
Question 1a. I'm concerned about maintaining the security of
Government data maintained and transmitted through mobile data storage
devices, particularly USB flash drive products. While I appreciate the
obvious day-to-day benefits of flash-drive technology, flash drives
infected with malware, as well as lost and stolen drives, present a
clear threat to our Government's information systems. I understand that
some flash drives use hardware--instead of software--authentication,
which protects the devide from malware and hacking.
Are you familiar with hardware-authenticated drives?
Answer. Response was not received at the time of publication.
Question 1b. If so, to what extent have you tested and evaluated
them?
Answer. Response was not received at the time of publication.
|
NEWSLETTER
|
|
Join the GlobalSecurity.org mailing list
|
|
|
