In order to guarantee the availability of the stored data, procedures on backup and restoration of software programs, data entered and operational data need to be in place. The following items need to be covered in the procedure(s):
- What is the frequency of backup?
- Verify retrieval of the backup before storing the backup (e.g., random checks).
- At least two generations of each backup set should be kept.
- There should be a system in place to manage the availability of the backup within an appropriate period of time.
- Stored backups should be checked for accessibility, durability and accuracy at a frequency appropriate for the storage medium (frequency to be specified in the SOP).
- Backup copies should be kept separate from the computerized system in a fire-protected area.
- Following changes to the system, change control should ensure the availability and integrity of the backup copies by restoring the data on a trial basis.
When to Test Backup & Restore?
The only real way to test back up and restore is to actually do it. The problem is that no one wants to do that with a live system. So what happens is that when lightning strikes and you actually have to replace a server and restore from backup, the restore does not go smoothly because no one has attempted to do this before.
Do this early in the project. Go through and do a complete Backup of the system and Restore to a new out of the box server. At FAT or early in SAT is a good time before any real GMP data is stored.
Take detailed notes so that you can draft an SOP that can be used when the lightning or fire or whatever destroys your server. Remember Ghost images are great if the new server is exactly like the old server. But usually 7 years down the road the specs of the new server will be different and the Ghost image will not work.
