By Gordon Jackson
published: Tuesday, March 18 2008
How IT's Always Been Done - A True Story
Since long before the dawn
of time, the
caretakers of silicon chips and magnetic media - the priests and priestesses of
IT - have dutifully served The Lines of Business. Dressed in the albs of the data center, these
white-smocked systems administrators, network managers and data architects faithfully
deployed hardware and software in response to the needs of The Lines of
Business. Mainframes, mini-computers,
micro-computers, rack-mounted servers, blades - more applications, more
business and more customers all meant more demand and more infrastructure, and
IT delivered. The Lines of Business
looked out and saw growth and said "It is good."
But all was not well in The
Enterprise, and the IT caretakers were summoned to explain to the Chiefs why so
much has been spent on infrastructure, yet so little is used; why is there so
much spent on power and cooling, yet the infrastructure appears to be idle? And why, despite these apparent wastes, is The
Enterprise being asked to purchase still more infrastructure? The IT caretakers could only nod and reply
that The Lines of Business require it so - since the earliest days of punched
cards and paper tapes and scheduled work it has been this way. It is how IT has always been done. IT serves The Lines of Business.
Pain Management
The current crisis didn't
just happen over night - it's not like IT (and Fiscal) executives woke up last
week and said to themselves, "Holy cow! That's a lot of hardware sitting around
not doing much of anything..." This
situation has been developing since the very first data center was built and has
been exacerbated by the combination of the Internet commerce explosion and steadily
decreasing hardware prices. The
pathology of the issue is the same, regardless of the nature of the business -
if there's a data center and an IT function supporting the business, then that
business is in all likelihood experiencing some or all of the negative
consequences stated above.
The hallmark symptom is
profound underutilization of server resources. Depending on your particular flavor of
analyst, the numbers are anywhere from 10-percent to 25-percent utilization in
aggregate. This isn't anything new to
you, and neither is the pain reliever that has been advertised as curing all
that ails you - virtualization. Server
virtualization, to be specific, however network, application and storage
virtualization are also seen as impacting overall system utilization for the
better. And for a time, the use of
virtualization will make the pain go away, or at least tolerable enough so as
to seem that the issue as been resolved. Addressing the symptoms though is not the same
as addressing the illness. You can make
the stomachache go away for a time, but the ulcer is still there until you
address it directly.
The symptom that
virtualization is addressing is underutilization, the problem, the real issue,
however, is the static allocation model that has been around since... (you
guessed it!) before the dawn of time. Until
The Enterprise (place your company name here) directly addresses this, IT will continue
to execute the way they always have, by statically allocating resources, both
physical and logical, to The Lines of Business. The only thing that virtualization
technologies change is the available supply of resources that the IT staff has
available to them. It is a temporal
solution and is really just pain management.
Out Goes the Status Quo
At this point, you might
well be asking yourself, "Alright then, what is the solution?"
In the same way that actually curing an ulcer
requires some radical lifestyle changes, addressing utilization, power
consumption and the static allocation model is going to require serious changes
to the status quo.
The first is that IT
will need to be embraced as a partner in The Enterprise and an actual provider
of service, quality of service that is. The
second, and this will be difficult, and likely will not happen immediately, is
that The Lines of Business will now contract with IT for specific qualities of service
for each of the applications and services that it intends to run, not for a
specific number of servers that IT needs to provision and manage on behalf of
that Line of Business.
The second requirement above
is the most important, not only to addressing the issues stated here so far,
but also to realizing the promises of Utility or Cloud computing, Software as a
Service (SaaS) and Service Oriented Architectures (SOA) and Service Oriented Virtualization
(SOV). It is the key, because as long as
The Lines of Business own (or at least pay for) the infrastructure, it will
always be statically allocated to specific applications - what I call the "I
paid for it, so it's mine" mentality. By
changing the status quo and putting the servers, and the processes by which
they are allocated to The Lines of Business, squarely into the hands of IT,
these resources can be dynamically allocated to the applications and services
that need them, when they need them and only for as long as they need them.
The result is an
infrastructure which is elastic to the needs of the business, intelligently expanding
and contracting around the demand for an application or service. This means an infrastructure that is
intelligently allocated to the applications and services based on the relative
importance of the application or the users of the application, as defined
within quality of service policies for these applications. The infrastructure is now more resilient and
efficient, as well as cost effective, because resources are only consumed when
needed and released when demand ebbs.
The rEvolution of the Data Center
Turning "the way IT has
always been done" upside down and changing the status quo are not trivial
undertakings by any means. Yet they are
necessary in order to achieve the benefits of delivering dynamically allocated
computing resources to enterprise applications based on business demands and
policies - a Real-Time Infrastructure. I've already mentioned the terms Utility and
Cloud computing, as well as SaaS and SOA. There has been a considerable amount already
written on each of these topics, and it is not my intention to repeat these
here, save to say that these are all subsets of the functionality offered by a
Real-Time Infrastructure. More
importantly, these are not achievable unless you are willing to fundamentally
alter the way you are conducting your IT operations: Real-Time Infrastructure is achieved through
evolution of the data center.
Evolution is simply the
change that occurs over time as a result of the pressures imposed by the
environment. Data center evolution, in
this case, is a necessary response to the increasing cost of energy for power
and cooling; the operational costs of managing increasingly complex
environments and the costs in lost revenues associated with failing to meet
consistent and high quality of service levels. Moving toward a Real-Time Infrastructure is a
direct response to these pressures.
Revolution is, at its
essence, a change of paradigm. Allocating
resources to applications and services intelligently, based upon importance and
in real-time response to unpredictable demands, is a complete breakaway from the
static, siloed way IT has always been done. Failing to do this will lead to the
evolution's flip-side: extinction.
¡Viva
la rEvolución!
Related Links:
Cloud Computing
Gordon serves as the Virtualization Evangelist for DataSynapse , and is an industry
veteran with twelve years of experience as a systems engineer, solution
architect and data center expert. In his role, Gordon is responsible for
communicating the value of application virtualization and real-time
infrastructure to various audiences, including customers, partners, prospects
and the market at large.
|
