Good backups

If one would ask which is the most important aspect of Vasudeva Server right now, I would say with no hesitation: backups.

No filesystem will provide automatic “back-up” options. Even systems like Apple’s TimeMachine are far from ideal, their «easyness» hiding their superficiality.

True back-ups protect the Archive from:

  • Failed media: hard drives die
  • Slowly-corrupting media: ie, drive spontaneously changing bits
  • Corrupted filesystems: structural corruption causing file loss
  • Accidental mistakes: delete a file by accident
  • Malicious destruction: virus, ill-intention agent
  • Whole-system failure: fire, floods, etc

A good, long-term back-up is performed regularly, respecting each of the following:

  • On a completely different filesystem,
  • on completely different media,
  • on a completely different machine,
  • in a completely different location.

Multiple sets should be stored in multiples countries.

Failed media, like losing a disk or two, can be mitigated by using a volume composed of redundant media, such as ZFS can offer.

Accidental mistakes such as accidentally deleting a file, can be protected against by using filesystem snapshots. ZFS is excellent in that.

Filesystem corruption, depending on the type of filesystem can try to mitigate structural corruption by intrinsic design or through features such as snapshots. Often this curruption is silent (bit-rot).

No filesystem can protect against whole-system failure, such as the entire machine dying in a fire, or a virus deleting files and snapshots.

Vasudeva Server needs museum-quality backup solution. Many professionals recommend Bacula. It may not matter which system we use at the beginning, as long as a real, true backup is in place.