In the last major IT buildout, we saw IT organizations adding
large numbers of small and medium-sized servers to host their
rapidly expanding IT application portfolios. During this time, the
best practices involved deploying a single application on each x86
server to avoid crashes or performance problems. However, this
approach led to very low utilization and higher support costs, the
most obvious of which are the acquisition costs associated with
needing to overprovision servers. Additionally, there are the
ongoing expenses; for example, a server operating at 10%
utilization still requires the same amount of power and cooling as
a server operating at 75% utilization. In the current state of the
market, for every $1.00 spent on new servers, the average
enterprise spends $0.50 on power and cooling. Further, for every
new $1.00 spent on infrastructure, the average enterprise spends
$8.00 on maintenance and operations, assuming a server to admin
ratio between 20 and 30 to 1.
To address these concerns, many enterprises have turned to
virtualization as a way of controlling costs and increasing the
flexibility of IT. If implemented correctly, the technology can
have spectacular results and reduce IT costs and improve
flexibility. The results of a recent IDC survey showed that most
customers that have deployed virtualization were able to reduce
their hardware costs by 20%. This reduction resulted in a savings
of, on average, 23% over the past 12 months due to lower hardware
costs, power and cooling savings, and real estate expense savings.
However, the survey also found that virtualization is not a simple
technology to deploy throughout an enterprise datacenter. The most
common problems that customers reported are associated with virtual
machine sprawl, network and storage problems due to the deployment
of virtualization, and changes in the underlying support and
management of the virtual machines.