With my increased focus on data center projects lately, it's interesting to see how business continuity and disaster recovery (DR) are a significant part of the equation and overall decision making. And that's a good thing!
In a previous tip, I've broadly discussed how your DR site should not be very close to your primary data center. A good rule of thumb is over 1,000 miles away, which pretty much isolates the DR site from a majority of the potential disaster events that could affect the primary site (e.g. power, weather, etc.). During data center consolidation projects -- for example, when reducing the number of data centers that a company has -- the location of the DR site becomes even more critical. Too often I have heard: "We'd like our DR site close by so our staff can get to it quickly." Unfortunately, that's the absolute wrong approach -- imagine the same geographic power outage, flood, or tornado affecting both sites simultaneously.
On the topic of staffing, the data center acquisition model also comes into play. When the DR site is positioned properly in a remote location, what should the staffing model look like? Does the company buy or lease the data center and put staff at that location? If so, it is important to consider how the data center site and associated staff will be used beyond DR. One option is running "dual production" data centers with each location serving as the DR site for the other; however, that becomes more complex as it is scaled up. Another option could be using the computing power at the DR site for development or test environments while accepting that these activities end immediately in a disaster (and during testing exercises).
Other data center acquisition options exist for a DR site including managed services agreements or co-location services through third parties. It's very easy to add a "skeleton staff on-demand" model to the contract, ensuring you have warm IT bodies as required. This is especially important if running the DR site as a "lights out" facility, or if other activities are required on a recurring basis (e.g., backup-tape shipping).
Speaking of tapes and backups, having an alternate data center close to your primary data center is not necessarily a bad thing as long as you accept that it's a business continuity site rather than DR data center site. With this concept, some interesting options can be considered, these include backing up directly to the alternate site at extended LAN or Fibre Channel speeds, and thereby having your backup tapes off-site once the backups complete. Geographic clustering and high-availability also become real options, especially if the alternate site is within synchronous data replication distance.
So, whether you are consolidating data centers, building new ones, or simply taking a hard look at your data center strategy, ensure business continuity and DR are high on your list.
ABOUT THE AUTHOR: Bill Peldzus is Vice President of Data Center and Business Continuity & Disaster Recovery Services, GlassHouse Technologies.