Nov 11 2010

Better Back That Data Up

When it comes to backups, make sure yours are done automatically, are tested regularly and have multiple processes.

Unfortunately, when you say “backup” to users, they hear “blah blah, nag nag.” Why? Because backup is boring. Users don’t care about backup.

The first rule of backups: If they aren’t done automatically, they aren’t done at all. Never rely on a user to do a backup, because both of you will be disappointed.

The only time users care about data loss is when they try to restore a file. Therefore, to users, backups aren’t important. Only restores are important. Because the goal of backup is a successful restore, the users are actually right on this. Techs focus on backups, but users only want restores.


The second rule of backups: Untested restoration procedures fail when they are most needed. When you conduct backups, you’re not finished unless you regularly test the restore success rate of those backups. The older the data, the more restore problems you will find. Luckily, most restoration sessions are for data less than two weeks old, so stay current.

Disk-based backup systems offer transfer speeds, random-access capabilities and easy offsite storage options beyond the reach of tape backup systems. While these systems make restoration faster, easier and more reliable, they must still be tested on a regular basis. Analysts say a full one-third of companies backing up to tape don’t test their backups. I’ll bet two-thirds of small companies backing up to disk trust the system too much and don’t regularly test their backups.

Another important reason to test file restoration is to ensure you’re backing up the proper data for each user. Desktop users might store all their data on a network-attached storage appliance or a file server, but what about notebook users? Mobile users tend to add their own applications, even when warned not to, and they may store that data in unusual places. Whether you use full-disk imaging for notebooks or just copy data files, test the restore process to a new notebook, and see if the user gets everything back. If he doesn’t, he isn’t backed up enough.

Where else do you store data? How about online applications? Hosted customer relationship management providers, for example, won’t lose your data, but users might erase the wrong files by accident (nearly half of all data loss is caused by user error). Can you recover the information if a user deletes “David’s Auto” rather than “Davis Auto”? Are you sure? Have you tested that recovery process?

The only time users care about data loss is when
they try to restore a file. Therefore, to users, backups aren’t important. Only restores are important.

Better Safe than Sorry

The third rule of backups: Almost every company needs multiple backup processes rather than a single procedure.

If your small business keeps all data inside devices on your local network, you might be able to get away with a single backup procedure. But mobile users, home users and remote users dominate the SMB world today, as only 25 percent of companies operate from a single location. More than one location for any of your files means you need more than one backup procedure. You need a backup system that covers all of your data files — no matter where the users and applications create and store those files.

While you’re thinking about all of these issues, rethink what you call your data protection process. We’ve established that users don’t like “backup.” But what if you instead roll out a series of “pre-restore processes” to protect your data?

Automatic, unattended, invisible “pre-restore processes” won’t aggravate your ­users nearly as much as “backups” will. And putting restore in the name puts the emphasis where it belongs: on restoring the data that’s critical to your company.

James E. Gaskin writes books, articles and jokes about technology from his Dallas-area home office. He also consults for those who don’t read his books and articles.