The Backup Conundrum: More Data in Less Time
IT managers are always looking for new ways to back up more data in less time. And even though there’s always room for improvement in backup speed, IT managers realize that performing a backup without disabling users would reduce the impact of the backup and therefore reach the same goal as improving backups with more speed. The problem continues to exist, though, with users still complaining that it takes too much time and too many resources to back up data. We decided to ask a few industry experts what advice they can offer in terms of solutions for achieving the goal of backing up more data in less time.
Point-in-Time Copies or Snapshots
One option is to perform backups from a mountable Snapshot, which means that the production data is never offline during a backup procedure. In other words, with mountable Snapshots, a backup window is no longer needed, according to Zophar Sante, vice president of marketing with SANRAD.
“Some storage solutions have the ability to take a Snapshot of a production volume,” says Sante. He explains that the Snapshot is a separate volume and is logically identified to the original production software. “A Snapshot volume does not contain a copy of all the actual data blocks, but rather is a virtual representation of the production volume.”
When a Snapshot is taken, he says, it instantly generates pointers to and protects all the blocks on the production volume. And when the backup software mounts the Snapshot volume and starts reading blocks for the backup process, it is actually being pointed back to the production volume where it reads the blocks and backs them up.
“The key to this,” says Sante, “is that the production volume is also simultaneously still accepting read and write commands without having to be frozen for backup. It keeps running, serving both client read and write commands as well as read commands from the Snapshot volume for the backup process.”
You may be wondering what happens to blocks on the production volume that is being overwritten while the backup process is taking place. The answer, according to Sante, is that the storage system knows that a production volume has an associated Snapshot volume. Before the old block is overwritten, he says, it is moved into the Snapshot volume.
The pointer that once pointed to the old block on the production volume is now pointed to the old block, which has been moved to the Snapshot volume. “So at the time the Snapshot is taken, all the blocks are mapped and protected,” he says. Sante sums it up by noting that you can keep all production volumes online and open 24x7, and take backups through the Snapshot volume without any “backup window.”