The CIO's Guide to Data Center Consolidation
For a long list of good reasons, many CIOs are consolidating data centers. A combination of best practices and powerful tools can ease the way.
By
John Zipperer
Environmental costs receive a high level of attention in Australia, even
affecting the way enterprises there pursue their data-center strategies. Just
ask Andrew Cresp, Head of Infrastructure, Technology Services, at
Insurance Australia Group (IAG), which underwrites some AU$7.4 billion
(roughly US$6.8 billion) a year in premiums in Australia, New
Zealand, the U.K. and Asia. Two years ago, the CEO of IAG, Michael
Hawker, decreed that the company would become carbon-neutral within five
years. Today Cresp's data centers are playing a major role in this effort.
Data centers everywhere are major consumers
of electric power, and that was certainly
the case at IAG. In fact, Cresp
discovered that the company's data center
in Melbourne, Australia, was consuming
more electricity than any of the company's
other offices—even more than its headquarters,
which houses some 3,000 employees.
The high energy usage helped point
out to the company's leadership that IT as
a whole was a big energy-user that it would
need to address if it were to achieve its goal
of significant energy reduction. "We had
the finger pointed at us for carbon usage
and carbon neutrality," says Cresp. To help
reduce data-center power requirements,
Cresp and his team at IAG have decided to
consolidate data centers as well as examine
why their carbon usage is so high, then take
steps to lower it.
Reducing power needs is just one reason
why enterprises in virtually all industries are
consolidating data centers. Other reasons
include assimilating systems from companies acquired in mergers and acquisitions;
cost-cutting; replacing outmoded server
hardware and software with newer, more
efficient models; and improving operational
efficiency.
In fact, improved operational efficiency
was the reason behind data center
consolidation that was cited by nearly
60 percent of experienced U.S. IT executives
and managers in a 2007 survey by
Forrester Research (see chart, p. 35).
Other reasons cited include reducing real
estate costs, reducing total cost of ownership
(TCO) and addressing cooling costs.
What they have in common is the desire
to make data centers more cost-efficient
and able to contribute more directly to a
firm's bottom line.
As Cresp notes, data centers make juicy
targets for efficiency initiatives. Because a
data center can be so costly to operate, the
fewer centers an enterprise has, the less
expensive it is to run that enterprise, even
when the cost of supersizing a consolidated
center is included.
At IAG, the data center was receiving
continually increasing demands from the
business, Cresp explains. Some 300 people
were working in the Melbourne data
center, which runs the technology for
IAG's 4,000 staff throughout Australia, as
well as for a division that handles private
lines of insurance.
Cresp realized he could do something
about the data center's exponential growth
by looking at server utilization. Before
the switch, his CPU utilization rate was
only about 5 percent. "When I saw that, I
was shocked, but then I found that it was
industry standard," says Cresp. "We
thought there was probably a smarter way
of doing this."