Why Power Management
for Servers?
Data centers use a lot of energy —
nearly 2% of the electricity consumed in
the United States, according to an EPA
report to Congress[1] - and recent
surveys[2] have shown that data center
energy is doubling every 5 to 7 years.
The increase in data center energy
consumed is due to both rising
computational demands and the
cooling costs for ensuring reliable
operation of computer systems.
Because servers are at the core of data
centers — and because the heat they
generate drives air conditioning costs —
they are a prime target for energy-
savings measures.
Not only are the computational
demands around the world increasing,
the power trends of volume servers
(those servers-1U, 2U, 4U and blades-
that are manufactured and sold in very
large quantities each year) are also
increasing. These trends were
documented in a 2012 ASHRAE
Datacom book[3] and are summarized in
Table 1. From 2010 to 2020 server
powers are showing dramatic increases
of 20 to 67%.
Performance of most systems is
constrained either thermally or by
power delivery. Every watt saved can be
converted into either increased
performance or lower energy
consumption. And it makes for a
greener planet. Deploying more energy-
efficient servers is a very effective
strategy
for
reducing
energy
consumption in the data center.
Reducing the energy consumption for
computation and cooling in servers is a
major challenge considering the data
center energy costs today. To ensure
energy efficient operation of servers in
data centers, the relationship among
computational power, temperature,
leakage power, and cooling power are a
significant focus of IT manufacturers.
What is Power
Management?
Simply put, power management is the
tradeoff between performance (how
quickly the work is done) vs energy (how
much power it takes to do the work).
In reality it is a very complicated,
multifaceted problem. And it is not just
about saving power! Instead, it is a
balance of optimizing performance
given multiple system constraints.
Some of the system features that can be
achieved are turbo mode allowing the
server to perform at much higher
performance. This can be achieved
since the components are actively
monitored and protected thermally and
for power delivery. On the other end of
the spectrum power can be limited
through a power capping feature which
in some cases can be activated by the
system operator. This feature might be
activated to prevent the data center
from exceeding the power capability of
the data center. Also, some features
allow various power saving modes to
meet
EnergyStar
requirements
necessary to sell some systems overseas
and to some governments.
48
7X24 MAGAZINE FALL 2014
by Roger Schmidt & Don Beaty, PE
Table 1 – Volume Server Power Trends to 2020
(fully configured, fully utilized max. load)
