What To Do When There's A Server Failure

There are few things in life you can say with no qualifications. Here's one: No one likes a server failure.

When a server goes down, businesses come crashing to a halt, hence the term "server crash." The words, "I think the server is down," sends a chill up the administrator's spine, as he or she knows the deluge of phone calls he is about to receive from impatient users.

The best defense is always a good offense. The surest way to avoid data loss is to include data recovery as a part of your overall backup and disaster planning. But sometimes, things just happen. So here are a couple of tips to follow now, so if (and when) something does go awry, fixing it will be a bit less of a chore. Send me more as you think of them.

Our first tip comes from Ontrack Data Recovery: Use a volume defragmenter regularly. Because a defragmenter moves the pieces of each file or folder to one location on the volume, each occupies a single, contiguous space on the disk drive. During a recovery, files and folders will be easier for data recovery specialists to find. Remember not to run defragmenter utilities on suspected bad drives; doing so could have damaging effects.

Secondly, remember to backup, backup, backup. Every night is ideal, but even more often may be needed depending on your business. But definitely do it daily, and also before and after major hardware or software changes.Third, manage expectations. Wayne Spivak, a frequent contributor to Server Pipeline, notes that by managing the disaster recovery process (from planning to actual recovery) efficiently, you will be in better shape to manage the impact when a server fails.

Fourth and finally, bear in mind that you can never have too many spare parts. Servers don't go down very often, but when they do, you never know what is going to break when they (try to) come back up. That tip comes from my friend and colleague technology writer David Strom, who has also worked as an IT consultant. He shared this anecdote: "I was working with a client that had a Tricord server many years ago. This server was running for years without failure, because it had redundant hot-swappable everything. We had to bring it down to replace a network card that we were upgrading. The server didn't start up, because the power cord that went into the machine from the wall was frayed and shorted out. Now, this machine was a very expensive unit, with dual power supplies, etc. But there was only a single power cord connecting the unit to the wall. We had to overnight a replacement from the depot."

So stop all that spring cleaning, and start planning for the worst. It's what I call the "Umbrella Principle": Take your umbrella with you, and it almost never rains. Make your plans for the "rain" now, and, hopefully, you'll rarely have to use them.