Backup Performance Art
When I first started working after graduating from George Washington University with a degree in International Affairs, I naturally became a Documentation Analyst at Central Bank and took a side job as the Night Auditor at the Hotel Jerome in Aspen. Just goes to show how unimportant one’s major can be.
At the Jerome, I added up the point-of-sale receipts and cash, then compared the totals with the sales recorded in the hotel’s central computer. I also changed the backup tape: a fresh tape went into the drive attached to the server, and the previous day’s tape went into the office safe.
While doing this, I would wonder how difficult it would be to restore the system. Had anyone ever tried? I didn’t know. My part was to change the tape and put the old one away.
A backup tape can survive a disaster without leaving you a working business. Someone still needs compatible equipment, the necessary software, and instructions for putting everything back together. The safe might protect the tape, depending on the safe and the disaster. It cannot supply the missing computer or the person who knows how to restore it.
The earthquakes that devastated Christchurch, New Zealand, in 2011 made disaster recovery an immediate concern for businesses. Losing access to a building could also mean losing access to computers, records, and backups stored on the premises. The tape in the office safe was of limited help if nobody could get into the office.1
We have a less dramatic version of this question at home. Our MacBook uses Apple’s Time Machine. We seldom test restoration. Time Machine takes much of the work out of making backups, which also makes it easy to stop thinking about what happens next. Could we recover an accidentally deleted folder? Could we get back to work on a replacement Mac? How long would it take?
Backup performance art is the reassuring routine of making backups without checking whether they can restore what we need.
The backup may be perfectly good. The trouble is that we have allowed the completion of a task to stand in for evidence that it serves its purpose. I changed the tape. The software reported success. We checked the box.
A modest test would tell us more. Restore a few files to a separate location and open them. Check that the important folders are included. Occasionally rehearse a larger recovery without disturbing the working system, and record what equipment, passwords, and instructions it requires.
For a manager, the useful question is: When did we last restore something, and what happened?
I still think that was a reasonable question for the night auditor.
-
The Paymasters Limited, “The Christchurch Earthquake – how we survived”, March 14, 2011. This contemporary business account describes the difficulty of reaching records and backups in buildings inside the cordon. It supports the point about access, rather than a claim that tape restoration itself failed. ↩