Accounting for Disaster Recovery in Data Center Consolidation
The folks at Forrester got back to me with some pretty relevant information about my last post, "Is Consolidating the Data Center at the Expense of Disaster Recovery?"
In that post I pondered this question: While consolidation is showing an increase, can companies that disperse failover capabilities between data centers afford the risk of single points of failure?
Analyst Stephanie Balaouras at Forrester has this to say:
Any company pursuing data center consolidation will still need at least two sites. Even with the most robust data center, a Tier III or Tier IV data center with backup power generators, UPS, redundant cooling, reinforced building integrity, redundant systems etc. in a low-risk location, it's impossible to mitigate all the risks. More than 85% of companies have at least one data center. What's changed from the past is that the second data center is no longer an idle recovery site; in most cases, companies use it to run secondary workloads like application development, testing etc. or it's another production data center, in which case each production data center can also act as recovery site for the other.
Server virtualization plays a major role in both data center consolidation and server consolidation. With server consolidation, you could view it as "putting more eggs in one basket," but actually server consolidation increases availability in a couple of ways. It's very simple to move virtual machines between physical servers networked to shared storage. Because the virtual machine images are stored on the shared storage, if one physical machine fails or it needs to be upgraded or it's end of life, you can move those VMs to an alternate physical server. When you combine this with storage-based replication to an alternate site, you now have the ability to quickly restart those VMs at the alternate site.
In fact, according to Forrester surveys, improving disaster recovery capabilities is actually the number one driver for server virtualization, it even edges out cost savings.
A meaty quote, indeed, but it gives us a lot more explanation as to how improving efficiency in disaster recovery may be accomplished. It also gives us a view to expand upon from the latest Computer Ecomomics research that was the catalyst for my questions and last post.
Baseline's David Strom says virtualization is the new clustering. From Strom's article:
The intersection of both technologies [clustering and virtualization] has produced a big business benefit that is widening the appeal: disaster-recovery protection. "In the past, you needed to buy another physical server in case the primary machine went down," says Bob Williamson, a senior vice president at Steeleye, a specialized virtualization vendor. "By using virtualization and hosting these servers at a remote location, enterprises can use the machines if their data center goes out. That lowers the entry cost for deploying wider-area disaster recovery and opens up this protection to a whole new set of companies that haven't been able to consider it before."
Considering the rise in virtualization adoption, this is making a whole lot of dots connect, and yet, three things come to mind: compliance laws, automation and reality.
Compliance depends on your industry, and for some of your companies, I have a hunch that the multi-data center failover game isn't going away as soon as you may desire. Automation implies monitoring, which, by Strom's account and his sources, isn't always foolproof.
A virtualization vendor and the researchers are saying it works, but I guess the question I would pose to you is, Is that actually the case? Is server virtualization working for you as a disaster recovery process as easily as the way it is being depicted?