Grant me the strength
to fight the things I can
change, the serenity to accept the things I can't, and the wisdom to
know the difference
| Type
of failure
|
Protection
|
Probability
|
Consequences
|
| Power
loss
|
Journaling
filesystem
|
easy
|
Very
Likely
|
Data
loss
|
| Disk
failure
|
RAID
|
easy
|
Fairly
likely
|
Great
data loss
downtime
|
| Data deletion (rm -rf /) |
Short-term backups,
snapshots
|
easy[1]/medium
|
Possible
|
Data loss
|
| Filesystem corruption
|
Long-term backups
|
hard
|
Unlikely[2] |
Data loss
|
| Bad blocks, slow
degradation
|
EVMS, future MD
|
hard (easy soon?)
|
Unlikely[3] |
Data loss
|
| Hardware failure
(excluding disks)
|
Redundant hardware
|
very hard
|
Fairly likely
|
Downtime
|
| Physical destruction
|
Off-site backups
|
very hard
|
Very unlikely[4]
|
Data loss
|
[1]Local snapshots,
versioning, and revision control offer protection from data deletion.
[2]Old established
filesystems are extensively tested, corruption unlikely unless using
experimental features.
[3]Most modern disks
will die from sudden failure (even accumulation of bad blocks manifests
itself as a sudden failure)
[4]Don't worry about it - or
worry, but know that worrying is as effective as trying to make the
Kessel run in a landspeeder.
|