Virtualization's Impact on IT Operations - Part One By Kevin Lees published: Wednesday, May 28 2008
Virtualization's Impact on IT Operations - Part 1
Virtualization continues to be recognized as
sufficiently mature for deployment in production environments. In fact, I would say it's
either rapidly approaching or has already arrived at the "knee in the curve." The
coming years will see an exponential increase in virtualized production
environments as it passes this bend and begins its, what I believe will be, rapid
climb up the steep part of the deployment curve.
What will influence how rapidly virtualization
technologies are deployed in production? Certainly its increasing technical
maturity will have a huge impact. For anyone who has attended VMworld or who takes
regular notice of virtualization industry announcements, there should
be little doubt that the advancements we see in virtualization's core
technology (for instance, hardware assisted virtualization) as well as in supporting solutions
(like VMware's Storage Motion, Stage Manager and Lifecycle Manager or Citrix's
XenCenter and XenMotion) will continue unabated. But, will the
maturity of virtualization's technologies and supporting solutions alone drive the
steepness of the deployment curve? If not, what else might influence
how rapidly virtualization is deployed in production environments? In my opinion,
it will be IT Operations. This series of articles will look at virtualization's
impact on IT Operations. This first article discusses the advantages virtualization
offers to IT Operations. The second article will address the IT Operations'
challenges presented by virtualization and the current state of available
tools to address these challenges. The final article in the series will consider
virtualization's impact on IT Operations from an ITIL perspective.
Today, IT Operations is primarily about Service Level
Management, well, and of course controlling the costs required to meet those
agreed to service levels. Pretty much everything IT Operations does it does to
maintain and improve the level of service it provides to other departments
within the company or externally to the company's customers. Given this, let's look at
how virtualization can positively impact service levels.
Virtualization can positively impact service levels
in several ways, some obvious, perhaps some not so obvious. The most obvious, and
most heavily promoted, is undoubtedly disaster recovery. Hardware independence
and the ability to run multiple virtual machines on a single target server,
even if production servers aren't virtualized, can make for faster recovery.
This is especially beneficial for companies that can't afford duplicate production
servers or rely on third party providers offering shared disaster recovery
facilities. If, on the other hand, production servers are virtualized, Recovery Point
and Recovery Time Objectives can potentially be improved through the use of
virtual machine snapshots, coupled, of course, with storage replication. All of
this leads to improved disaster recovery service levels.
While disaster recovery is one of the most accepted
and most widely written about advantages of virtualization for IT Operations
relative to improved service levels, what I really want to explore are perhaps
some less obvious, but potentially as impactful, applications of
virtualization. One of the most exciting, coming from someone who was involved in IT Operations
management at an application service provider, addresses the process
that should ideally exist after a software release exits test and development (or a
new application or application update is brought in from a third party)
but before it's deployed into production - the operations' readiness process.
The operations' readiness process allows IT
Operations' personnel to familiarize themselves with, test and certify the production
deployment readiness of a software release (as in operating system patch, new
third-party application/application update, or internal release)
in a production-like environment. Therein lies the rub. More often than
not, hardware costs preclude the required, production-like environment (Operations'
Readiness Lab) needed to test readiness with any effectiveness. In an
x86-based production environment virtualization makes creating the Operations'
Readiness Lab much more affordable, and hence much more realizable. Use of a
virtualized Operations Readiness Lab will increase your rate of success when
releasing software into production and lower the risk of planned downtime
extending beyond that specified in a Service Level Agreement.
While on the subject of deploying software releases
into production, how relieved will your administrators be, in a virtualized
production environment, when a release into production involves deploying a
previously tested and certified (via your recently created virtualized Operations
Readiness Lab) virtual machine? Not to mention if, heaven forbid, there is a problem
with the release in production and "rolling back" involves restarting the previous
virtual machine instead of backing the release out of a physical production
server? I have, unfortunately, been through releases that lasted six hours on a
Friday night instead of two due to having to back out a release. I, for one, believe
the relief the administrators, not to mention myself, would have felt if we were
faced with restarting the original virtual machine instead of backing the release out of
a physical server would have been palpable - not to mention how it would have
relieved the pressure on planned downtime for the month!
Another area of positive impact on IT Operations for
which virtualization shows promise is zero-downtime backups. Basically, the
challenge addressed by zerodowntime backups is one in which an application,
predominately a database, must be quiesced prior to backup in order to avoid
data loss due to write requests being dropped. In order to ensure that all
write requests are flushed prior to the backup, the application must be taken
offline. Currently, one tool, at least, begins to address this. VMware's Consolidated
Backup provides for virtual machine activity to be captured in a redo log while a
snapshot of the virtual machine is taken. Once the snapshot is complete, the redo
log is "played" to the running virtual machine ensuring that no virtual
machine activity was lost while the snapshot was being made. This sequence of events
provides zero-downtime backup of the virtual machine. Granted, this may be a
stretch for a database of any consequence, for which placement in a virtual
machine is debatable anyway, hence why I wrote virtualization holds promise for
zero-downtime backups. As virtualization improves, though, with technologies
such as hardware-assisted virtualization, I can foresee a time when
zero-downtime backups of databases in a virtual machine could become a reality for IT
Operations'.
One final area I want to touch on, in which
virtualization has a positive impact on IT Operations, is application deployment in response
to changing business needs.
To stay competitive, successful businesses must be
able to rapidly respond to changing market conditions or, depending on the
business, a customer request. Traditionally, a gating factor for how quickly IT Operations
could respond to a business unit's need for rolling out a new
application in response to a customer request, a competitor or a market change was hardware
provisioning. Today, by making use of virtual machine "golden templates", a
functioning virtual machine, certified at least to the operating system level, can
be up and running in 15 to 30 minutes. Now, this in itself presents both an upside
and a downside to IT Operations, the potential downside being virtual
machine sprawl but I'll save that for the second article in this series where I'll look
at the challenges virtualization presents to IT Operations.
The virtualization advantages I describe in this
article are based on my experience in pre-virtualization IT Operations at an
ASP. With the exception of the Operations Readiness Lab, which we were fortunate
enough to have a physical version of, my team would have directly
benefitted from the positive impact of a virtualized environment. Meeting the
planned downtime component of our customer Service Level Agreements would have
definitely been less stressful. But, virtualization also presents
challenges to IT Operations. Fortunately, tools are and continue to become
available to begin addressing these challenges. I'll look at this aspect of
virtualization's impact on IT Operations in the next
article in this series.
Related Links:
PMP'n Part One, PMP'n Part Two
Kevin is the principal consultant at Premier Project Management, LLC,
where he specializes in IT infrastructure architecture development as
well as planning and managing IT infrastructure, virtualization, and
data center consolidation/relocation projects. He is a Project
Management Professional and VMware Certified Professional with 26+
years of technical, management, and consulting experience in systems
integration, project management and IT operations. Recent engagements
include performing an enterprise-wide IT Infrastructure &
Operations assessment as well as planning and managing a
multi-datacenter consolidation / relocation, using virtualization, in
the publishing industry; managing the implementation and operational
"go-live" of two e-mail platforms for an international e-mail ASP; and
providing technical project management and virtualization services for
the assessment phase of a multi-datacenter consolidation / relocation
project in the on-line, e-mail marketing service provider space. Kevin
can be reached at
This e-mail address is being protected from spam bots, you need JavaScript enabled to view it
This e-mail address is being protected from spam bots, you need JavaScript enabled to view it
.
|