How frequently should a disaster recovery test be conducted? And should it be determined by changes to your infrastructure?
Depends what you mean by "test." Formal test events entail a lot of activities that planners would be smart to design into the strategies themselves. We should be able to validate data protection and recovery on an ad hoc basis -- every day!
With geoclustering, for example, you can simulate or perform failovers to remote sites on a simulated or actual basis at any time of the day or night. That leaves logistics testing, which should be performed two to four times a year, and most certainly following significant changes to business processes or infrastructure.
Tabletop exercises and plan walkthroughs are great testing vehicles that can be used much more frequently than formal off-site tests. They have the merit of being much less expensive to conduct, they have fewer logistical requirements, and they give teams a chance to interact with each other as they walk through documented procedures for accomplishing their work.
Just keep in mind that a disaster recovery test also must be conducted in an objective-driven way. You might not be testing to a stopwatch, but interactions need to be kept sufficiently formal and professional to exercise the concept of a strategy or procedure in an effective way.
This was first published in August 2013