Page 84 -

C h a p t e r 3 : h a p t e r 3 : M 55 55
M i n i m i z i n g P o w e r U s a g e i n i m i z i n g P o w e r U s a g e
C

Servers
You can reduce the amount of energy your servers use by deploying blade servers and by
virtualizing your servers.
Blades are entire computers contained on a card that can be inserted into a larger device.
As such, rather than one server taking an entire rack, 20 blades can be installed into one unit.
Usage
Blade servers consume about 10 percent less power than equivalent rack mount servers
because multiple servers share common power supplies, cooling fans, and other components.
Blades are popular because they not only reduce the amount of space needed, as
Figure 3-4 shows, but also because they use less power.

NOTE Currently, five manufacturers account for more than 75 percent of all the blade servers on PART II
the market. These manufacturers are HP, IBM, Sun, Fujitsu/Siemens, and Dell.
Many organizations have a large number of servers, each running one application. This
separates and isolates the units, and if there is a failure only that application is affected.
Although there is some logic behind this setup, the reality is that each server only has a 5 to
10 percent CPU utilization rate. This boils down to a lot of hardware taking up a lot of
space, and not doing as much as it should.

NOTE In 2007, Symantec conducted a study that asked organizations if they were considering
server consolidation or virtualization. Of respondents, 51 percent said they were considering
consolidation, and 47 percent said they were considering virtualization.
Case Study
Ask.com has taken a unique tack in its efforts to maximize what its servers do and what it
gets from them in return. The company asked its server vendor, Dell, to build servers that are
customized for its very specific functions. That is, each server is built with specific memory,
processors, disk space, and power supplies, all tuned to the application the server supports.

Multiple servers can be Blade servers
consolidated and virtualized
onto a single blade server.

FIGURE 3-4 The work of many servers can be offloaded onto one server containing many blades.