Backup Policy
Backup Policy
TIG-managed AFS and NFS storage space1 is backed up according to the following schedule:
Backup Schedule
Full backups
- Taken upon creation and population
- Mirrored tape writes
- One copy stored offsite for one year
Monthly backups
- Taken monthly
- Mirrored tape writes
- One copy stored offsite for one year
Daily backups (incremental)
- Taken daily
- Archived for one month
Backup Granularity and Recovery
The monthly backup is the only level of granularity guaranteed over time. For recovery requests:
- Less than two weeks old: Single-day resolution usually available
- 30-40+ days old: Monthly-level resolution only
We archive backup tapes as long as possible, but cannot guarantee physical tape integrity beyond one year or the availability of drives to read legacy media.
Snapshots
AFS storage is snapshotted nightly. A copy-on-write snapshot of each volume is available; only the AFS home directory has the snapshot mounted by default (as ~/.snapshot).
If you need the snapshot feature enabled for your AFS project directory or AFS group directory, email help@csail.mit.edu.
Three-Level Backup System
This backup approach with mirrored full backups provides several benefits:
- Conserves tape storage (full data written once; monthlies are ~20% of full size)
- Protects against single tape loss (mirrored monthlies mean no data loss from one failure)
- Enables recovery from lost full backups using previous monthly and full tapes
- Maintains two copies of each monthly backup
-
Except in the case of hyper-volatile shared filesystems, such as /data/scratch. ↩︎


