Planning for Failure

Abstract

Planning for failure is one of the most important things you can do as a DBA. However, you should keep in mind that planning does little good unless you actually test your plans. This is true whether you are talking about backup and recovery, failover clustering, database mirroring, or any other techniques you are using to attain high availability. Having a DBA on staff is like having an insurance policy. Unfortunately, most of the time your phone rings, it’s not a user telling you things are going great and to keep up the good work; they only call you when something goes wrong. A crisis or high-stress situation is not the time to realize that a process executes differently than what you originally expected. People will be asking you a lot of questions, and you will be much better off if you already know the answers. You should know what constitutes a failover, how long a failover will take, what happens in different failover scenarios, and what to expect after a failover has occurred. As long as you are well prepared, you will be able to remain calm and make the right decisions when a failure does arise.