Cisco Cisco UCS C210 M1 General-Purpose Rack-Mount Server Livre blanc
Page
IT & DATA MANAGEMENT RESEARCH,
INDUSTRY ANALYSIS & CONSULTING
Data Center Management: The Key Ingredient for Reducing Server Power while Increasing Data Center Capacity
©2010 Enterprise Management Associates, Inc. All Rights Reserved. | www.enterprisemanagement.com
To help understand data center power efficiency, the Power Usage Effectiveness (PUE) metric gauges
the ratio of total power used by the data center against the power delivered to computing equipment.
In other words, PUE evaluates the overall efficiencies of the data center from a power distribution
and utilization perspective. A PUE of 1.0, for example, is a perfect score, indicating that each watt
consumed by the data center translates into a watt delivered to the equipment (which is obviously unat-
tainable). Another consideration is the newly unveiled Energy Star rating for data centers, introduced
by the U.S. EPA in June, 2010. The new rating, which is entirely based on PUE, is awarded to data
centers in the top 25% of their peers for PUE, and is audited by a third party. Organizations that may
wish to attain the new Energy Star rating will definitely want to do everything possible to improve their
PUE rating.
As one might expect, factors such as heating and cooling inefficiencies, as well as switches and vampiric
As one might expect, factors such as heating and cooling inefficiencies, as well as switches and vampiric
loss to equipment like uninterruptible power supplies (UPS) and power distribution units (PDUs)
directly affect the PUE. Today’s state-of-the-art data centers may attain PUEs of 1.8 or lower, but most
legacy data centers report PUEs between 2.0 and 2.4 (or even higher). PUE is important because it
illustrates the hidden cost of data center power consumption. It can also incent departments to con-
duct thermal, cooling and power delivery studies, identifying problem areas and pinpointing candidates
for increased power efficiencies.
Servers comprise a major percentage of overall data center power usage, and studies have shown that a
Servers comprise a major percentage of overall data center power usage, and studies have shown that a
watt generated at a power plant diminishes to 0.30 watts by the time it arrives at the data center, drop-
ping to only 0.17 watt by the time a server translates it into business value. Reversing the math, every
watt saved at the server saves ~1.8 watts of data center consumption
3
and obviating the requirement to
generate ~5.9 watts in the first place.
4
Maximizing performance per watt should be a key goal.
One factor to consider when comparing server vendors is that today, virtually all major manufacturers
utilize the same Intel Xeon processors that leverage the same Nehalem microarchitecture. This simpli-
fies power comparisons between vendors, as servers that use the same processor type and speed levels
the playing field, moving the conversation to other differentiators, including system architecture, other
hardware design considerations, and management efficiencies.
In the next section, we perform a detailed comparison of blade servers from HP and Cisco, both of
In the next section, we perform a detailed comparison of blade servers from HP and Cisco, both of
whom use Intel Xeon/Nehalem processors. First, the management capabilities of both vendors are
examined, and then a quantitative analysis of the power requirements for both vendors is presented.
Cisco and HP: Comparing Blade Server Power and
Management Capabilities
It is interesting to note that the price of server hardware, as a percentage of overall TCO (total cost
of ownership), has decreased steadily over the past four years. Server hardware acquisition cost as a
percentage of total TCO has been dropping as technology has advanced, and is now approaching only
20% of total three-year TCO.
Moreover, processor price as a function of TCO is now relatively constant across all vendors, with the
Moreover, processor price as a function of TCO is now relatively constant across all vendors, with the
processor price itself comprising only a very small, insignificant percentage of the total solution cost.
With CPU costs basically equal and overall hardware costs steadily declining, the major TCO differ-
3
0.30 watts / 0.17 watts = 1.76 watts
4
1.00 watt / 0.17 watts = 5.88 watts