By Ram Shanmugam, Sr. Director of Product Management
Unless you’ve been living under a rock for the last half-decade, you know that virtualization is changing the landscape of IT and data centers. In terms of financial impact, virtualization untethers applications from physical servers, creating valuable savings. In terms of disaster recovery impact, virtualization makes recovering applications easier – MUCH easier. It’s as easy as copying a file to a computer and running it. Here’s the kicker: the world is not 100% virtualized yet. Data centers are becoming increasingly virtualized, but most data centers today are still some part physical and some part virtual. That is to say, they are “hybrid” environments (to support my point, Gartner told us in a recent inquiry that they estimate 50% of all workloads today to be running on virtual machines). That means, 50% are not.
While newer applications are being run on exclusively virtual workloads, there are still plenty of mission-critical apps running on a combination of mainframes, Windows servers, Linux/Unix systems, and virtual machines. Given this scenario of a hybrid production environment, the challenge for CIOs becomes: “How do you best protect and recover applications within a hybrid infrastructure within certain recovery time objectives (RTOs) and recovery point objectives (RPOs)?” Or, in other words, “How do you think about Disaster Recovery in this new semi-virtualized world?”
Well, here’s my short answer: as long as we are living in this hybrid world, virtualization is an added layer of complexity that requires some adjustments to your recovery strategy and infrastructure. Most DR fundamental principles don’t change, but a few tweaks are required. I will elaborate upon these in this blog and in two more blog posts to come.
What Doesn’t Change
- Application tiering. Applications still need to be tiered according to their respective cost of downtime. You should still assign an RTO and RPO to each application based on its overall impact to your business.
- You still need to move your data from your production environment into a recovery environment (some might call this a “DR site” or “secondary site.”) How you choose to move the data is dependent upon the RTO and RPO that you assigned above.
- You still need to ensure compatibility between production and recovery environments. After all, if you let the infrastructures and technologies between the two sites diverge too much, how can you use one to recover the other?
What Needs Tweaking
Since your primary site is now a hodgepodge of physical and virtual (meaning multiple applications running on multiple platforms, multiple hypervisors, and multiple storage), you should expect that your recovery site will be the same as well. If you’re doing DR yourself (we call this the “self-insured” model), then you’ll need to ensure the total compatibility of your physical and virtual compute layers between your primary and secondary sites. The “tweak” I am referring to is the addition of the “virtual” layer, with all of its attendant hardware, software, and people/expertise.
I’ll be back later this week to spell out more about each point above. Stay tuned!
 *Gartner, Inc., Top Five Trends for x86 Server Virtualization, Thomas J. Bittman, March 22, 2012.