Dedupe To Tape

Deduplication to tape

(26 October 2009)

Running deduplication on your filer is one thing; backing up deduplication to tape in deduplicated form is another.

The primary concern is data integrity. With standard backups, loss of some blocks of the backup set can be fatal to particular files. Thanks to deduplication, loss of some blocks can now be fatal to many files.

When performing backups, your number one priority is your ability to restore.

(Yes, I know that there is an implied "within an acceptable time window", but for me that is a business decision, not a technical one. If a business is willing to pay to have the ability to restore $X in $Y, great. Absent that, data integrity is everything. If you can't restore it, it really does not matter how fast you can't restore it.)

One way deduplication might be advantageous is if you run multiple copies of your backup tapes. For example, if you back up to disk, and then spin two copies of of the backup set to separate tapes, that might make the risk worth while.