My first encounter
with a data center was a trip down to the basement of our office building,
where the DEC PDP-11 sat
ensconced behind a big glass wall: a
mass of tape drives, processors and peripherals that was connected to the dumb
terminals sitting back in our offices. I
further date myself by adding that it accepted punch
Not too long after
the PDP-11 was retired, all kinds of crazy things started happening. Servers appeared in lab rooms, dedicated to specific
projects. Some departments had their
own IT networks. In many cases, companies had no idea about how
many IT assets they had, and where they were located.
A great example of this
is the US government’s Federal Data
Center Consolidation Initiative, (which has since been updated by the Data Center Optimization Initiative). When
this initiative started in 2010 with the objective of reducing the number of
data centers by 40 percent by the year 2015, the initial tally of data centers stood
at 2,094. At the
end of 2015, the estimated number stood at 11,700,
an increase of 459 percent instead of a reduction of 40 percent. This
increase was due to new builds as well as an expanded definition of a data
center that counts servers and telecom closets as small as 100 ft2. It’s estimated that more than 70 percent of these
are server closets.
government was not alone in the efforts to consolidate; many companies also had
data center sprawl due to expansion, acquisitions and in many cases, lack of
management of IT assets. There are many
cases of massive consolidation, such as HP’s
initiative to move from 85 data centers down to 6 massive data centers,
saving $1 billion in the process.
Similarly, with the advent of what were initially centralized cloud data
centers built in areas where power was cheap and available; it appeared that
the pendulum had swung back to the centralized model for good.
This may have been
the case if it hadn’t been for the sustained growth in bandwidth, ever
increasing mobile networks and the development of new applications that require
low latency. These trends lead to the
emergence of edge networks and the need for edge data centers. While centralized data centers can help
reduce costs, they also drive up latency, which can severely degrade
time-sensitive applications like 4K video,
gaming and augmented reality.
Similarly, transport costs can be significantly reduced by deploying and
edge data center with local content, rather than pay to transport it from a
centralized data center. And finally, with the onslaught of devices that make
up the internet of things, processing must be done in real-time to truly reap
the benefits of the IoT. This can be
accomplished with an edge data center.
And while data center
consolidation will continue based on the savings around combining disparate
facilities, edge data centers will also be built: in existing central offices, in collocation
data centers and in other locations closer to end users.
The data center
continues to evolve, and it is likely that future phases will follow.
If you are attending
Datacloud Europe 2016, June
8-9, in Monaco, you can see more data center solutions from CommScope at booth