Within the Cloud: Technical Concepts of Cloud Computing - Deploying and Managing a Cloud Infrastructure - page 78

consume about 16 million kilowatt hours a year, which is roughly equivalent to the consump-
tion of 1,400 average U.S. households per year.
In the planning stage, engineers typically look at the energy efficiency of each piece of
IT equipment that goes into the data center. This might be an energy-saving measure, but
it would not be enough. We always need to look at the bigger picture, and in this case we
have to consider the end-to-end power efficiency of the entire data center system, from the
power grid's point of entry into the facility to the equipment that uses this power. We have
to look beyond energy-efficient servers and virtualization and start looking at our power
infrastructure as well.
Consider that for each power conditioner and converter that electricity passes through,
some electricity is lost. Each power module has its own efficiency rating, so only by adding
the efficiency rating of all the modules that the power has to go through to get to the serv-
ers will we see the true energy efficiency of the entire data center system.
A 2008 study by the Eaton Corporation found that overall system power efficiency can
be optimized by finding the total efficiency from end to end. The study showed that stan-
dard 480 V AC power distribution systems used in many data centers are inadequate when it
comes to power efficiency. The study compared the newer, more-energy-efficient 400 V AC
and 600 V AC power distribution systems. Eaton concluded that the 400 V AC variant would
be more efficient and would save more money in the long run compared to the 600 V AC
alternative. When you pair that with high-efficiency UPSs and servers, you can be sure that
your data center is optimized for power efficiency.
But that study was done in 2008, and newer data centers have better alternatives for
power efficiency that also conforms to state regulatory laws. Our conclusion here is that
when calculating for efficiency and optimization, you have to see the whole picture and look
from end to end, not just at the equipment you are familiar with. As cloud computing profes-
sionals, we must concern ourselves with efficiency and savings wherever we might get it.
The Cloud Infrastructure
Though cloud data centers and traditional data centers have similar hardware, their infra-
structure design, or how everything is tied together, is different. In fact, depending on the
purpose of the cloud data center, its infrastructure design may differ from another cloud
data center. Again, all of this is down to what type of services are going to be offered or
provided, and it all has to do with optimization.
Even with many differences in infrastructure design and components, cloud computing
infrastructure has some common areas:
Distributed File Systems Cloud computing often relies on distributed file systems, which
are spread across multiple hard drives and machines or even across multiple data centers.
They provide speed and reliability as well as data redundancy in case of failure. Data is
never stored in just one place, so when a unit fails, another will simply take its place and
carry the load or provide the required data. This offers high availability to the user, yet it's