Every now and then, I like to take time and concentrate on tape backup and its role in Disaster recovery (DR). While so much attention is placed on disk backup, real-time and snapshot replication, and other more modern technologies, tape backup allows for both a good basic plan and a vital addition to a more well-rounded enterprise-class DR solution. My point is that you shouldn't put all your eggs in one basket when it comes to protecting your data. These backup methods have advantages and disadvantages that you should take into account when creating your recovery plan. Relying on any single method could put you at risk, while mixing elements of available technologies adds to your security.
As a methodology for DR in itself, tape allows you to take a copy of your data and move it to another location besides the server it's currently sitting on. This means that on a regular schedule — once per day, for example — you would commit all data or just the changed files to a removable tape cartridge. At that point, the tapes should be removed to another location for safe-keeping, as anything that hits the server room (fire, flood, etc.) could endanger any on-site tapes as well.
From this basic idea, you can begin to craft even more complex solutions using just tape backup. Many organisations perform daily backups, full weekly backups, and sometimes even monthly backups. Some tapes are kept on-site for immediate restoration in the event of need, while others are rotated off-site for safety. Organisations can tape once per day, take snapshots with different tools, and commit the snapshots to tape on a regular basis throughout the day, etc. This venerable technology still has a place as basic DR in its own right.
Once your organisation grows beyond the ability of tape to provide full DR, you will still want to use tape technology as part of your overall DR plan. Replication systems leave you open to the possibility that corrupted data on one server could easily be replicated to other servers before you detect the malicious intent of an intruder. Snapshots and other point-in-time copies are still housed on moving disks — great for quick recovery in some situations, but ultimately undependable in a full-blown disaster. Since the disk must reside in a server for the snapshots to take place, you cannot effectively use a snapshot system to natively get data away from production machines. While you could replicate the data to another site, that data would still fall prey to malicious intent and/or virus activities, which both could easily destroy data on the production and replicated servers alike. Adding tape into the equation allows you to easily remove point-in-time copies from volatile locations and house them securely at a remote site for true DR.
Tape copies that have been securely warehoused off-site can help you recover from disasters that these other methodologies do not adequately protect against, and are therefore still a vital part of your solution.
No matter what size organisation you are part of, or how intricate your DR plan is, you will still want to make tape backups a part of your overall plan. Even if you hope you never have to use them, they will more often than not save your data when — literally — all else fails.
