The correct answer to the interview question, "Have you ever had to restore using a backup?"

I'm constantly amazed by the answers DBA candidates give to the question, "When was the last time you had to restore using a backup?" The most common is, "Our environment is pretty stable/robust/well engineered, we've never had to do that!"

I assume it's meant to indirectly demonstrate their capabilities as a DBA, engineer or architect, and it's great that you've never had a production situation that required you to recover.

But what happens when you do have to recover.

Recently, a candidate told me that they hadn't had to recover but the process was well documented.

Right.

I wondered, "When?" Documents can be incomplete. Processes change over time. Systems are upgraded and improved and migrated to new hardware. That documentation may have been accurate when written, but the evolution of the environment may have made it confusing, misleading or even wrong. It may have made sense to the person who wrote it but it may contain assumptions or be missing steps. You may not want to discover any of this with the CEO standing over your shoulder.

Recovery is arguably the most stressful situation a DBA may encounter in their career. A slow-performing database is still running, but when you're down waiting for recovery to complete, all eyes are on the DBA. A multitude of variables add to the pressure—do I have all the files I need, did I remember to set the environment correctly, is there enough space on the recovery system?

Most DBAs who haven't had to perform a production recovery don't understand that the situation is rarely ideal. Databases, like kids and pets, only get sick after the doctor's office has closed for a long holiday weekend. (The most significant recovery I've performed began late on the Friday of Memorial Day weekend. I was alerted as I was walking out the door for a trip with my wife and kids.)

Consider that during a restore you may be at the mercy of others. You may have to rely on storage admins to make space available, network admins to open routes, system admins to build hosts from bare metal, colocation facilities to grant physical and virtual access...

Why would you leave any of that to chance?

The correct answer to the question, "Have you ever had to restore using a backup?" is (minimally) "Yes, during the last quarter I performed a restore as part of a regular readiness exercise."

If your employer doesn't require this, fight to make it their policy. At the very least you'll be able to say that you take backup and recovery seriously and demonstrate a commitment to keeping an environment stable and highly available.

DBAs that aren't practicing recovery are akin to soccer or hockey goalies who never practice defending penalty shots. The heat of the moment is not the time to be figuring it out!