This article was submitted by
Greg
Ness
In the same way that the rise of the Internet led to the
enterprise web (web-enabled enterprise applications) the public and
private cloud is driving IT to new modes of operation, most of
which demand more agility, more capacity and an even keener
focus on operating and capital expenses. This profound
tension -between a substantial evolution in demands on IT and
an extended period of economic doldrums- has placed CIOs in the
crossfire.
The Good News
Companies like VMware, Microsoft and Citrix have helped to ease
infrastructure challenges with new automation and management
capabilities, centered on servers, virtualization and the promise
of greater IT automation. The public cloud players (Amazon,
Google and Microsoft) have also helped enterprises with edge cases
where agility was needed or building for peak capacity wasn't
economically feasible.
The Not So Good News
Networking and security players have been notably slower in
addressing today's agility and scale-related networking challenges;
network automation has progressed more slowly than server
automation as many network vendors have continued to focus on
hardware-centric and feudalistic product and marketing strategies.
The hope for private clouds is the evolution of the network and the
emergence of cloud operating systems which can increase efficiency
and agility by managing and securing workloads beyond VLAN
constraints.
Yet the biggest challenge facing CIOs today may have more to do
with the data center facility itself.
The Coming Data Center Race
The last ten years have seen an accelerating pace of change
within IT, especially when it comes to agility, capacity and the
need for capital and operating efficiency. Yet most data
centers are obsolete, and many enterprises are in the dark at what
point obsolescence occurs.
Most data center decision makers have been focused on securing
adequate floor space, walls and back-up generator capacity, and
have otherwise treated the data center as just another building in
a portfolio of real estate holdings. They have no idea how
energy efficient their data centers are under most if not all
operating conditions and measure capacity by available space.
Virtualization has eased this problem by enabling more power
density (more server capacity per rack) but that has shifted the
issue to efficient power capacity. Increased growth means
increased waste as a data center passes its point of efficient
power and cooling capacity. Yet many operators don't have a
clear idea of where that point is beyond whether or not the data
center is full of racks.
This has left many CIOs in the precarious position of driving
innovation from facilities where growth may risk putting an IT team
at a disadvantage.
The data center industry and the pace of innovation is at the
core of the problem. Data centers need to be more scalable
and more energy efficient than typical office buildings because
they: 1) are disproportionate consumers of power; 2) are
increasingly strategic to a company's operating posture; and 3) can
be the most significant factor in a company's ongoing ability to
profitably innovate and grow.
Yet most data centers are designed (by today's standards) to
become obsolete in less than ten years, forcing more transactions
than necessary for the same amount of IT capabilities and driving
up capital and operating expense. They leave the "at
risk" CIO tasked with increasing agility and capacity at the
expense of rising operating and capital expenses, which exacerbates
an already difficult challenge.
For example, in ten years a highly efficient and vertically
scalable data center can reduce operating expenses by about $70
million and allow for a doubling of IT capacity within a single
building. Those who deliver IT services and apps from a more
efficient data center can outperform those in obsolete facilities
on pretty much every quantifiable measure.
The Good News: The Race is On
This widening gap between obsolescence and increasing demands on
IT has spurred data center innovation in two areas. At the
low end of the market a multitude of vendors are offering highly
efficient containerized data centers, complete with racks, which
can be deployed quickly. They offer viable alternatives to
retail colocation and public cloud, but do have their drawbacks.
For example, you can still run out of space quickly and you
might end up paying more for the contents than you could otherwise
obtain directly from a supplier. Some of them are not
comfortable work environments for your IT teams and many offer a
fixed offering that may not address your unique infrastructure
needs.
Above 500 kWs of power consumption CIOs may want their teams to
have more control over power densities, electrical and mechanical
architectures, amenities, etc. That leads to a larger,
wholesale data center, where enterprises can have virtually
complete control over a building financed by a public or private
REIT.
These REITs, however, can vary widely in terms of design and
efficiency innovation. Some have a preset "one size fits all"
design (your capacity grows you simply buy/lease more space) while
others offer comprehensive customization and collaboration options,
including the vertical scalability game-changer which allows for
considerable opex savings.
You can see examples of highly efficient and vertically scalable
facilities by looking at Facebook's Prineville (especially for
infrastructure innovation), and Vantage's Santa Clara Campus for
facilities innovation. Note: I work for Vantage Data Centers, a
private wholesale data center REIT.
