Backups are not archives

We are pretty anal about backups, but that does not mean that we can reproduce your project at any point in time. The goal of backups is to be able to restore that filesystem to its state just before the calamity struck. That is quite different from being able to recreate the analysis in the paper that you submitted a year ago and are just now hearing about from a cranky referee.

If you wish to be able to reproduce your work (as many scientists do) consider using a version control software such as git While many people use github, 'git' the opensource software that powers github is flexible and can be used on our servers in a very sleek command line way.

If git is not your thing, you can also use 'tar' to make archives of an entire directory tree -- for example just after submitting a manuscript for publication.

What we do to preserve your data


We take "snapshots" of all data directories -- including /90days every day and we retain them for 7 days. We also keep Sunday's snapshots around for a month. Snapshots are easy to comb through and copy from. The only limitations are (1) snapshots are taken only once a day so if the thing you want to recover was not on the system overnight, there is a good chance that it will not have been captured by the morning's snapshot. (2) If you don't notice the thing that's missing or damaged until a week later -- you could be out of luck. Definitely ask though as we can sometimes do magic.


Our backups are kept in the clouds where we hope never to use them. They do not extend as far back in time as our snapshots so they're only going to come in handy if something really really bad happens.