Archiving Disasters

“But we have a short-term recovery plan in place surely we can get rid of tape backups now?”

This is a statement I hear time and time again and the answer (as always) is “it depends.”

The key issue is the difference between disaster-recovery and archiving, but first some key questions: What type of data do you want to be able to recovery? (files? a database? an entire OS?)  How long can the data be unavailable? (never? 1 day? 1 week?)  How much data can you lose? (no data? 4 hours? one day?

If you need database archive and recovery, can your database solution leverage a bare-metal or file-level restore? Or is a traditional client-based database-aware solution required? In my experience, database recovery is only needed in the short term, so while an off-site copy is required for DR a archival history of the database is not.

Disaster recovery implies a complete recovery (OS, apps, data) in a short period of time – no history, just get the system back the way it was before the problem.  Archive implies partial recovery (only relevant files) from a specific point in the past. There are certaintly solutions that combine the two solutions, the hurdle is cost. 

Completely recovering from a disaster requires more storage space (OS, temp files, applications) than an archive solution that only saves the most critical files.   The good news is that storage space is getting cheaper, a 1.5TB LTO5 tape is about $80 (plus tape drive!) and an external 1.5TB WesternDigital USB drive is about $80.  So the possilbility exists (if the DR solution supports it) to keep DR data for archive purposes at only $53/TB.

DR solutions are also geard towards complete recovery and simply getting back one file might not be easy.  However, recovering a file server when you only have the files means installing, patching and securing an OS, then restoring the files.

Archive solutions (primarily tape) are also the traditional backup solution and are often already in place.  Administrators know and understand tape backups and when it comes to security a known trusted solution is hard to abandon.

The results of these trade-offs usually leaves companies supporting multiple solutions, one solution aimed at quickly restoring an entire server (DR), one aimed at keeping snaphots-in-time of their data (Archive), and one to support database-level DR.

This multi-faceted approach is my currently preferred solution for most clients and can be accomplished a variety of ways including storage array snapshots or (in a virtual world) VMware snapshots for short-term complete recovery and traditional backup to tape (with off-site storage) for longer-term recovery.

The database solution needs to be discussed with the application vendor and DBA to assure a viable recovery. For some resources (Active Directory, Exchange) the ability to replicate off-site may be built into the product, for other additional 3rd party resources or application licensing may be needed.

This entry was posted in Disaster Recovery and tagged , , , . Bookmark the permalink.

Leave a Reply