Many of us were shocked the other week that a company as prominent as Danger, Inc. could make such a rookie mistake by losing the data of T-Mobile’s Sidekick customers. As a system administrator, if there is one thing you absolutely have to get right, it’s backups. Here are six ways to botch them.
1. Relying only on RAID
So you’ve made a smart decision by making sure that your company’s data is on redundant disks. Disk arrays using RAID 1,5 and 6 can continue to function if a drive fails. Great, but what if you lose multiple drives due to a power surge, defective controller, fire, flood, or user error? What if the data becomes corrupt or is accidentally deleted? RAID is great for uptime, but it isn’t even close to being a complete backup.
2. Relying only on online media
Perhaps you’re taking advantage of the plethora of cheap, spacious external drives to backup your system. That’s actually not a horrible idea if afterward you disconnect the drive and move it to an alternate location. However, keeping that backup online and connected is a bad idea. Imagine that your system becomes compromised by a virus or a hacker; all data on all connected drives could easily be erased. What if your power supply fries and it sends out a jolt that kills both internal and external drives? Keeping your backup hard disk away from you system minimizes the risk of a single problem wiping out all your data.
3. Keeping all backups on-site
Many of the things that could cause you to need your backups are the very things that can destroy your on-site backups. Nature can be cruel, and data closets with lots of electronics are excellent candidates for a fire. Shortly after 9/11/2001, I heard an anecdote of a company that operated from one tower and kept its off-site backups in the second tower. Obviously losing its data wasn’t the worst of this company’s problems, but losing all your data is a quick way to lose your business.
4. Not testing your backups
When was the last time you performed a test restore of your data? If the answer is never, then how do you know your backups are good? How do you know that you’re even backing up the right data?
5. Not making multiple backups of your data
It’s true that most media is pretty reliable. However, disks, tapes, and optical media can all become damaged or corrupted. Performing regular backups and rotating your media are good ways of making sure that a single bad tape won’t ruin your business.
6. Not backing up your system before a significant change
If you’re performing an operating system update, major software upgrade, or hardware upgrade, you’d better backup your data before making the change. Performing any sort of significant update is just the sort of excuse your system needs to corrupt its databases or become unbootable. It’s best to be prepared by first performing a backup even if your database is clustered. Just ask that sysadmin at Danger who’s probably looking for a career change.
(Although, in hindsight, it must be painfully obvious that it’s not a great idea to trust your data to a company named Danger.)
Michael Scalisi is an IT manager based in Alameda, California.