Mirror, Mirror, on the Tape...

Still confused about what Stac's Replica does that's different from what Cheyenne's ArcServe does...or from what other backup systems do? Here's a quick guide to popular backup options.

File by file: With traditional file-by-file backup, each file is copied as a single, named entity, regardless of whether it's compressed. This approach makes finding individual files quick and easy, but creating thousands of individual files on a tape in the first place can be t ime consuming. An analogy can be drawn to a word processing file containing many individual words that can be found and copied without any problem.

Disk mirroring: Traditionally, mirroring s oftware makes an exact copy of a disk, sector by sector. This is easy and fast to do on the backup end. But when it comes to restoring a file, it's almost impossible to find any individual item inside that image quickly. In terms of our word processing analogy, this is like sending a fax. Instead of having a document that's easy to search, you end up with one huge image file that's nearly impossible to search automatically. Worse, the document includes large amounts of white space that you have to transmit and navigate around to get to the words you need. Like image backup, it's easy to handle, but it's almost impossible to find anything within it quickly.

Data mirroring: What Stac does with Replica, in effect, is to combine the best elements of both approaches. Replica makes disk images that don't include the time-wasting white space of ordinary mirroring. Getting to an individual file still isn't as fast as when working with a file-by-file copy, but it's far faster than dealing with an ordinary mirror-style backup. And note, too, that when we say "data mirroring," "data" includes application code.