We held a backup workshop that was really relaxed. We all sat around the campfire for a roundtable of:

  • what people expected of the workshop
  • what their backup strategy was (if any)
  • worst backup story

I came out with this set of notes about people's expectations, the software they use, and some good practices.

Expectations

What people were looking for coming to this workshop:

  • learn best practices in professionnal world
  • new developments in backup software
  • a good, one tool to backup my computer(s)
  • encryption
  • share and discover pain points

Software

This is the software that was described in the workshop, in no particular order.

However, we have good confidence in Back in time, dejadup, rsync, bup and attic (now borg) being or becoming great software options. mediagoblin is also cool but is not backup software per se.

  • Back in time
    • Useful for desktops with a GUI
    • uses rsync and hardlinks
    • incremental
    • history: can go back to previous versions easily
    • supports multiple profiles
    • plus it has an awesome theme song
  • rdiff-backup
    • incremental
    • history
    • metadata support (permissions, etc)
    • used on many servers
    • old (last release in 2009!)
    • not very scalable
    • commandline
    • Python
  • rsync
    • commandline
    • incremental
    • no history
    • metadata support requires root on both ends
    • very widely used
    • well maintained
    • often used as the basis for other scrips, for example rsnapshot
    • C
  • dropbox
    • proprietary service
    • basic example of a "cloud" service, where you store a copy of your files on a third party server
  • "copying folders"
    • simply copying important folders to external drive regularly
    • file managers have various levels of support for this, some will merge, other will just replace
    • ideally they could use rsync or similar...
  • symform
    • also cloud
    • also proprietary
    • p2p - get 1GB free for every 2GB shared
    • syncs files between machines
  • dejadup
    • based on duplicity
    • incremental
    • history
    • GUI
    • encryption
  • bup
    • incremental
    • history
    • deduplication
    • commandline
    • can't remove old backups
    • very fast
    • may be difficult to use
    • very solid
    • based on git
  • attic (now borg)
    • incremental
    • history
    • deduplication
    • commandline
    • encryption
    • very fast
  • mediagoblin
    • free software gallery for video, audio, photos, 3d models...
    • not a backup tool, but a free service for cloud alternatives

There are of course many more backup software alternatives, but those were all mentionned in the workshop.

Some software added after the workshop:

Strategies

Here are all the strategies that were described in the working group.

  • squirrel: spread your little files everywhere, but you may have trouble finding them
  • diaspora: a little similar, share your files with everyone
  • external drive: backup to an external drive
  • offsite drive: swap that drive with another one in another location (see also JWZ's backup documentation)
  • encrypted drives: same, but use Full Disk Encryption
  • usb key: all important files on a single USB key that is backed up at home whenever it is inserted at a certain time (requires a cronjob)
  • cloud: copy regularly your important files on a cloud provider like Dropbox
  • mailing flash cards: store crucial files on a flash card and send it by email
  • outsourcing: have your friend, partner or workplace do backups for you
  • don't backup anything: you are ready to loose everything on that device (maybe it's just a copy from the cloud!)... and you will.

As you can see, there is a huge variety of backup strategies and that is just from 6 people!

Horror stories

Those are the basic horror stories you can expect from having no backups, or having backups break:

  • backup stolen, lost a year of photos without backups
  • lost everything
  • ubuntu's default sucks, doesn't do anything
  • lost 600 emails
  • physically destroyed own computer
  • destroyed a whole client virtual machine
  • forgot crypto password

What to backup

A question that was asked in the workshop is how to decide what to backup. Here are the answers we came up with:

  • the work you don't want to have to redo all over again
  • non-public, personnal data
  • copy of your passport? notice how we hold our digital data so dear while we don't backup our paper data!
  • your passwords
  • choose what not to backup instead of selecting what seems important, that way you're certain you won't miss anything

Good practices

And here are some random advice on top of all that:

  • automated: backups should be automated, otherwise you will forget to do them
  • have reminders to verify the backups, do test recoveries, otherwise you won't know that it is possible to restore them
  • backup frequency: duration of data you are ready to loose
  • backups can be noisy at home: modulate backup sounds so they become art!