Skip to content
Martin Pool edited this page Feb 7, 2014 · 2 revisions

What emergencies is a backup system actually supposed to preserve you from? How would Conserve cope with each of them?

See for example this Google tech talk on backups:

  • There are a lot more bugs in code than there are asteroid impacts.
  • Merely storing multiple copies doesn't protect against bugs or operator errors.
  • Local copies don't protect against site outages
  • Diversity in storage technology (hardware and software) protects against bugs
  • Have to actually restore them, to disk, using the real restore path

Malicious deletion of data

Someone breaks into your account, and deletes (or encrypts) the files.

Key point here is that if they own your main machine they may very likely be able to get access to other machines or accounts where backup data is stored.

One defense is to have backups stored on media that is physically disconnected, but that is in tension with it being fresh.

Might be nice if there was a way to remotely store onto a machine that allowed appending and retrieving but not remote deletion. Or, make the remote machine the master (which is an option in bup.)

Disk failure

Accidental deletion of files

Typically, only some of them - so fast random access is important.

Application-level corruption

Clone this wiki locally