| May 26, 2005 By Sonny Discini
Organizations are constantly looking for new
ways to cut costs, improve efficiency and
provide better services. You'll find that IT
departments, specifically, are constantly
involved in this process because of the
rapid introduction, evolution and
termination of technologies. While the
potential of reducing costs across a shared
infrastructure is appealing as a means of
lowering the capital expenses associated
with acquiring technology, the real economic
benefits are realized by streamlining and
lowering the time and expense associated
with managing change.
One of the newest fronts in streamlining
and efficiency has been the adoption of
virtual machine environments. Those who
deploy VM instances tout that they save
costs by not having to buy extra hardware,
rack space and ultimately the cost of
personnel to maintain the hardware that is
normally associated with huge clusters of
servers.
Additional benefits include snapshot
capabilities for rolling back servers to
pre-installation state. That is, should you
install a patch or an app and the uninstall
does not go well, the VM instance can be
"rolled back" and the admin will not have to
hunt and peck for leftover installation
components. Software developers and quality
assurance groups also praise VM environments
for the ability to quickly setup test
environments utilizing minimal physical
hardware, and once testing is over, those
environments can be sanitized quickly for
the next round of testing.
Let's look at one popular VM developer,
VMware, which has been very successful in
getting their product out to large
corporations. Many of the benefits listed
above are expounded upon by a large
customer, 7-11 Inc. Matt Ramseyer, a Senior
Business Analyst for 7-Eleven says, "VMware
does for Intel servers what Henry Ford did
for the automobile. In the same way more
people were able to afford a Model-T because
of the innovation of the assembly line,
VMware enables us to allocate more server
resources to developers. It has been a
pleasant change now that getting a server up
and running for a new project is not the
delay it once was."
Back to top
Ramseyer goes on to say, "We get an
exponential cost saving because of the way
VMware pools server resources to optimize
hardware utilization. A hosting company
without a VMware solution couldn't compete
with what we could do with VMware software."
When Matt obtained an evaluation copy of
VMware GSX Server, he quickly realized
benefits that he hadn't expected, running
five virtual machines on a single physical
system.
"All we really needed to do at the time
was evaluate how different browsers affected
our Web sites," says Ramseyer. "Because
hardware cost was an issue, we needed to
figure out how to run multiple operating
systems on a single physical machine. Once
we started to evaluate VMware software, we
realized that being able to quickly deploy
servers helped development teams get their
jobs done."
In May of 2004, 7-Eleven added VMware
VirtualCenter and VMotion technology to its
virtual infrastructure for optimal server
administration, managing servers centrally,
providing new servers and moving virtual
machines to different physical servers as
resource needs dictate.
"We just move the virtual machines around
when we need to," says Ramseyer. "Because
all of the virtual machines are centrally
managed by VMware VirtualCenter, we can
allocate a server whatever resources it
needs at the time it needs it."
Surely these are strong arguments for
deploying VMs but are there any downfalls?
You'd better believe it, and they aren't
unique to a single VM vendor.
Since security and administration are
dynamically opposing forces, you can imagine
that when you have such a dramatic change in
administration, you're going to have an
equal and opposite shift in your security
stance.
Back to top
Patch Management
The Achilles heel in most current VM
environments is that they depend on a host
OS. Let's say that you have a single server
that hosts 20 virtual servers. Patching the
20 virtual servers is no different than
patching ordinary non-VM servers, however,
what happens when you need to patch the host
OS?
You guessed it. All of your VM servers
must be taken down when you are forced to
reboot the host OS. This means that core
services for an entire division may be
affected because the host OS has to be
rebooted after a patch is applied. Adding
additional stress to the situation, given
the frequency of patches that come out of
Redmond Washington, before long, management
will become less tolerant when you tell them
that the entire sales division will be down
while you reboot the host OS and restart all
of the VM servers.
Separation of Sensitive Information
Smaller shops that do not have the
funding to purchase all of the hardware
required to engineer a network that provides
isolation of sensitive data may fall victim
to a common bad VM practice - adding VMs of
servers that contain sensitive data on the
same physical host that contains all other
servers with data of varying sensitivity.
The security implications of this are
critical when you take into account that
attackers have already begun fortifying
their toolboxes with the necessary tools to
exploit VM environments. Techniques range
from searching for VM tools and drivers all
the way to mini apps that detect the
presence of VM environments.
Back to top
Virtual Support
Microsoft still maintains their position
on support within VMs with the exception, of
course, of their own VM products. A
Microsoft knowledge base article states,
"Microsoft does not test or support
Microsoft software running in conjunction
with non-Microsoft hardware virtualization
software. For Microsoft customers who do not
have a Premier-level support agreement,
Microsoft will require the issue to be
reproduced independently from the
non-Microsoft hardware virtualization
software. Where the issue is confirmed to be
unrelated to the non-Microsoft hardware
virtualization software, Microsoft will
support its software in a manner that is
consistent with support provided when that
software is not running in conjunction with
non-Microsoft hardware virtualization
software."
Now, don't think that premier level
support customers are any better off. The
folks paying a considerable amount more than
most will simply be subject to terms similar
to Symantec's in that they will make an
effort but may require you to reproduce
issues outside of the VM. This can take a
lot of time and effort considering all the
variables that have to be addressed when
trying to isolate a VM issue. If you've ever
been in the middle of a support issue that
could be the cause of more than one vendor,
you understand the pain associated with this
task.
Considering that support contracts can
cost hundreds of thousands of dollars and
downtime costs have no limit, it would be
extremely wise to check with your vendors
before deploying production VM architectures
otherwise you may end up voiding all of your
support contracts. Generally speaking, this
isn't a very good way to maintain gainful
employment.
Are the risks worth the rewards?
Without question, VMs are here to stay
and there are plenty of reliable statistics
that prove cost savings. However, will these
savings be justified when a published
exploit is released that threatens virtual
environments? When reading the product
benefits listed by VM vendors,
interestingly, you never see improved
security as one of them. Like any other
security saying, it's what you don't see
that you have to worry about. Given this and
lackluster vendor support one has to ask
whether it makes sense to jump into the VM
game this early or wait until VM vendors get
a better handle on securing VM technology.
Back to top
|