|
|
Backup Failures
|
- Loss of one incremental file is disastrous/fatal,
- Subsequent incremental backups might not contain some of those contents
- Drive vs Tape
- tape backups require somebody to put new tapes into the drive/changers
- disk backups can be and should be 100% automated
- for simplicity, let's assume the disk and tape backup process/strategy/scripts is the same
- Backup failed due to system and methodology failures
- failures apply to tape or disk backups
- partition goes bad after reset/power failures...
- somebody didnt check the backup system regularly/randomly
- somebody didnt change the tape yesterday
- its just backup... it doesnt matter that a backup file is missing
- incremental backups was NOT properly implemented
- network connectivity went down
- dns died
- nfs timeout due to some other pc went offline
- ethernet cable got pulled
- fan died
- flaky memory
- power supply surge
- 110vac power failure
- cpu failure
- incompatible tar/compressed formats
- etc..
- Disk Failures
- 100% disk full ( on the backup system )
- disk crashes...
- disk cable insulation was slowly being eaten away...
- Tape Failures
- Tape failures are subject to more fatal ADDITIONAL tape failures
- Tape can ge 100% full too
- Tape can corrupted/bad
- Tape drives can bad
- Tape heads need to be cleaned
- Tapes take too long to backup and/or restore ( cannot do rdiff )
- Tape cable insulation was slowly being eaten away...
- "tape folks" go on vacation, call in sick ...
- "tape folks" forget to change the tape ...
- Raid5 merely protects you against ONE disk failure..
- it is NOT a backup
- it will NOT help you if you lose 2 or most disks out of 5 disks in the raid5 array
|
|