Transforming capacity planning in enterprise data centers

When capacity is locked into technology silos, servers
assigned to some applications might use just a small
fraction of their processing power. These servers take
up valuable floor space and drive up power and cool -
ing costs. And all of these challenges are intensified
when IT assets are often spread across multiple and
varied data centers.
The use of virtualization technologies can help you
break down the silos and consolidate applications and
workloads onto a smaller number of servers, each of
which is better utilized. But to really make the most of
your investments, you need to have a better view of
your server workloads, available processing power,
the utilization of your virtual and physical environment,
and server energy consumption on an ongoing basis.
What’s more, you need a better view of your opportu -
nities to increase server utilization and improve appli -
cation performance by shifting workloads to servers
with free capacity.
Ideally, today’s common approaches to capacity plan -
ning should help you in these efforts. And they do, but
only to a limited extent.
Current approaches to capacity
planning
If you help keep an enterprise data center up and
running, you have undoubtedly wrestled with capacity
planning questions. How should we go about consoli -
dating legacy systems? How can we re-balance
workloads to improve application performance and
take greater advantage of our available processing
power? Where should we stack individual workloads
within our existing environment?
Elsewhere within the enterprise, executive managers
may be asking a complementary set of questions. How
can we gain more value from our substantial server
investments? How can we control rising data center
costs while still maintaining an infrastructure that will
support business growth? How can we increase the
agility of our business technology to enable new
business initiatives and drive better business outcomes?
To help answer questions like these, organizations turn
to capacity planning. In some cases, data centers have
teams of dedicated experts who use complex software
tools to analyze capacity needs, usually for a limited
number of mission-critical applications. In other cases,
server administrators engage in an ad hoc process
that uses spreadsheets and guesswork to develop esti -
mates of the capacity needs of various applications.
When a formal capacity planning program is in place,
a planner might produce one capacity plan per week,
focusing on a single server running a single applica -
tion. That plan will most likely be devoted to a mission-
critical application. Applications that are less important
to the business might never be part of a capacity plan.
In the absence of detailed and definitive information
on historical and current server workloads, utilization
levels and power usage, capacity planners and server
administrators often have little choice but to over-
provision resources just to be sure that systems can
meet the peak loads of certain applications. So even
with a dedicated capacity planning team in place,
capacity waste becomes a part of day-to-day business.
2