Whilst there are many inherent problems associated with backup, one thing that tends to work out quite well is that there's usually a good correlation between data that gets frequently accessed, and data that's important.
Since the availability of frequently accessed data is generally assured, when one migrates between systems, the frequently accessed data is most likely to be copied over (rather than moved over). This means that there will be redundant backups of the frequently access data, and due to the correlation, the important data.
- Summary:
- 1) Data you use most often will be copied to new machines
- 2) It'll be put in easy to reach places
- 3) Therefore, it backs itself up
- Compare WuWei
One unfortunate factor that William Loughborough pointed out to me is that "used" is often conflated with "important". When I was explaining Freenet to someone years ago, which stores the most frequently accessed documents on more nodes, they pointed out that much of the valuable archival material from the past is only very very rarely accessed, and yet is much more important than much modern dreck. I didn't have an answer as to how that should be handled!
Type: Archival Pattern