Article Written by Aaron Rallo,
founder and CEO of TSO Logic
The terminology itself implies
that everything related to virtualization - including its costs - is virtual,
not real. And because they are easy to deploy, many have considered VMs to be a
"free" alternative to physical machines. But in reality, virtual machines carry
a broad array of capital and operating expenses, starting with the acquisition
of physical servers needed to support them down to the cost of licenses and
supporting storage infrastructure. To get a handle on the true costs, it's
important for you to have a complete understanding of the hardware and software
resources required to deliver applications.
Start
with the basics.
Measuring
VM host density can go a long way towards determining the cost of compute, in both
on-premise and cloud environments. But until recently, the lack of good, solid
data has made it difficult to know what density levels could be tolerated
without impacting end users. Routinely, IT organizations deploy compute without
ever looking at how it is used or even if it is actually being used at all. By leveraging
intelligence from an analytical engine that ties compute back to applications
and their owners, you can measure the cost per application alongside
utilization levels to ensure you not only have cost visibility, but can also
see where there's waste. From here, you can effectively optimize application
delivery, while meeting service level agreements and decreasing costs.
Key measurements reveal the true costs.
As virtual environments grow, they
become more difficult to measure and manage. Diving in to get relevant data
means you have to dig through old sources, talk to developers who may no longer
be around, and try to piece things together. To move away from these manual
processes, an analytics and BI layer can be used to discover applications,
compute, and utilization levels while also making recommended/automated actions.
Once you have visibility, key cost metrics can be determined, including: What
is the cost of a guest? How does that cost change based on hardware type? What
are the costs per active vs. idle guests? Are there zombie VMs that we can
remove or run in less expensive environments? These are all pieces of data to
consider when transforming your data center or moving to the cloud.
Waste
not, want not.
Virtual machines
have some very visible costs. They take up space on hosts and in storage and drive
up licensing fees. Then there are hidden costs, of which underutilization is a
key culprit. In our day-to-day work, we typically find that about
one out of every three virtual servers is comatose. So if one-third of your VMs
is doing nothing at all, don't go out and buy more licenses, hosts, and
compute. An easy way to cut costs is to cut out the waste and repurpose your
do-nothing VMs or get rid of them altogether.
The
key to success.
Over
the past several years, efforts to virtualize have clearly reduced the amount
of physical infrastructure required but has led to significant amounts of
sprawl. The next generation of savings will be driven by insights into how virtual
compute is being used and how it can be optimized. Going blindly into the cloud
or needlessly deploying more VMs will drive up your costs. So be sure to measure
and evaluate your virtual environment just as you do all other aspects of your
business. It's not too late to get started, and your business will thank you
for it.
##
About the Author
Aaron Rallo is founder and
chief executive officer of TSO Logic, a global company that enables IT
transformation by delivering unprecedented visibility into data centers and the
cloud. www.tsologic.com
