Gene Kim is one of the original authors of Tripwire,
a software product used to manage configurations and change. Gene is
willing to share his thoughts on virtualization with the Security
Laboratory thought leadership series, and we certainly thank him for
his time!
Gene, let's start with change, how important is change management?
Stephen, every day, as information security practitioners, we live with
the reality that they are a single change away from a security breach
that could result in front page news, brand damage, or regulatory
fines. These issues are clearly not confined to security, but they
impact business at the highest level. Consequently, security
practitioners strive to implement IT controls to mitigate the risk of
fraud, loss of confidential customer information, disruption of
critical business services and data integrity, inaccurate financial
reporting, and the list goes on.
Change also creates risk from an operational perspective - every IT
organization lives with the daily reality that they’re always one
change away from an outage, a catastrophic episode of unplanned work,
or something that causes audit or security issues. All of which
jeopardizes the completion of planned work, which is what they’re
supposed to be working on.
But certainly we can't stop change, Gene.
So true - in fact, it seems like it’s just part of the human
condition: change happens, and the pace always seems to be get faster
and faster.
This has some serious security implications, though. The need to
respond quickly to urgent business needs makes it more and more
difficult to effectively balance risk and controls. Most business
functions now require IT in order to conduct operations. In fact,
almost every business decision requires at least one change by IT - a
trend that continues to grow.
So how does virtualization factor into this discussion? It is
clearly hot, according to an article on SecurityFocus, "Intel and AMD
are building support for virtualization into their CPUs to make the
technology easier to implement and faster to run."[1] What is the
driver for it, being green?
That’s definitely part of the reason, Stephen. Other reasons are
the need for increased agility and the ever increasing cost and
complexity of IT. All of these have contributed to the rapid adoption
of virtualization technologies.
Virtualization makes it possible to build and deploy IT releases and
changes into production faster and more economically than ever before.
So it is a brave new world, what does it mean for security? Is
virtualization going to be an agent for or against security? An
Infoworld article by Tom Yeger points out, "Multiple virtual machines
sharing one physical system are likely to use a sequential range of IP
addresses, and they often have identical local administrator passwords.
Crack one, and you’ve cracked all servers with similar
characteristics."[2] What is your sense on the security of virtual
machines?
Some virtualization experts claim that virtualized computing
environments are fundamentally no less secure than physical computing
environments. Others claim that virtualization can enable better
security. Both of these claims can be correct, but only under certain
conditions.
The reality is that when information security controls are improperly
implemented or neglected in virtualized environments, real security
risks and exposures are created faster than ever.
Haha. I think the fear is that virtualization can create a scenario of Unsafe at Any Speed: the Designed-In Dangers of the American Automobile
by Ralph Nader.[3] Of course, I’m exaggerating to make a
point. Virtualization can be very secure or very insecure. But what is
definitely true is that what was safe at 60 miles per hour may not be
safe at 200 miles per hour, which is the faster pace that
virtualization enables. And this is the potential dark side of
virtualization: that the information security controls that adequately
controlled risks before virtualization may no longer suffice.
Virtualization enables rapid deployment, potentially allowing insecure
IT infrastructure to be deployed throughout the organization faster
than ever. The unfortunate truth is that the people who deploy this
infrastructure often circumvent existing security and compliance
controls when doing so. Unfortunately, the risk these deployments
introduce is only discovered when a security breach occurs, an audit
finding is made, or the organization loses confidential data or
critical functionality.
How popular is virtualization, how many organizations are already using it?
For better or for worse, virtualization is here. Tripwire surveyed 219
IT organizations and found that 85 percent were already using
virtualization, with half of the remaining organizations planning to
use virtualization in the near future. Furthermore, VMware found that
85 percent of their customers are using virtualization for
mission-critical production services. In other words, inadequate
information security controls may already be jeopardizing critical IT
services with risk introduced by virtualization.
There seem to be two keys to information assurance, to
configure systems properly in the first place and to detect anomalous
traffic.[4] How important is configuration in the virtual world?
Most information security practitioners now attribute the majority of
security failures to misconfiguration resulting from human error.
According to Gartner, "the security issues related to vulnerability and
configuration management get worse, not better, when virtualized."
Also, according to Gartner, "Like their physical counterparts, most
security vulnerabilities will be introduced through misconfiguration
and mismanagement."[5]
Why? Among other reasons, insecure virtual server images can be
replicated far more easily than before, and once deployed, require
great effort to discover and bring back to a known and trusted state.
Analysts have published some startling predictions on these information
security implications: Gartner predicts that "Through 2009, 60
percent of production VMs will be less secure than their physical
counterparts" and that "30 percent of deployments [will be associated]
with a VM-related security incident."[6]
The good news is that it doesn’t have to be this way.
Where do people make their big mistakes?
The security risks occur primarily at two levels: at the virtual
machine manager (VMM) layer where the host OS resides, and at the
virtual machine instance layer where the guest OSes
reside. Misconfiguration can occur in both layers that would allow
security risks to be uncontrolled and unmitigated.
The fact is that when done manually, setting configurations properly is
tedious, slow and error-prone. As information security practitioners,
our goal should be to ensure that all configuration settings at the VMM
and guest OS layers are properly defined, implemented and verified.
There’s already lots of great guidance on how to do this from
respected third parties and vendors, including Center for Internet Security, VMware, and so forth.
And of course, because we are appropriately paranoid, we must "trust,
but verify." This is where you need automated tools to help
achieve and maintain known and trusted states, so you can find variance
and quickly fix it. Information security will own parts of these
settings, but where they don’t, they need to hold the relevant
parties accountable for ensuring that their portions of the
infrastructure are locked down (e.g., VMM manager, servers, networks,
databases, applications, etc.)
Incidentally, information security can’t do any of this
they’re not aware that virtualization is being used. This
requires some situational awareness, so some sleuthing around may be
required to even find out where virtualization is being used, and by
whom.
Thanks for taking the time to share with us and contributing to
the thought leadership series on the security laboratory Gene and
congratulations of the birth of your first son Reid, he looks just like
you, though a tad smaller! I can't wait to see how long it takes till
you give him his first composition book!