Why we have disk quotas

To guard against disk failures, simple user errors (e.g., deleting an important file by accident), and other risks, we make backups of all user home directories. Our backups consist of snapshots taken every 4 hours, as well as weekly tape backups. There are, of course, costs associated with making backups: either disk costs, tape costs, or simply the time needed to make the backups. And these costs make it necessary to limit how much data you can store in your home directory.

We try very hard to make the quota limits large enough for normal CS-related work. Unfortunately we can not offer unlimited storage space in your home directory. As a compromise, we all have scratch directories (e.g., /scratch/dduck1 for user dduck1), which are not backed up. Your scratch directory should be accessible from any lab machine, and is a great place to store large datasets and software downloaded from the internet.

You can link files and directories that are in /scratch back to your home directory, if needed. This makes them seem like they are in your home directory, but they’re really on /scratch, and don’t count against your quota. Using ln like this should work (assuming user dduck1 wants to link largefile from their scratch directory to their home directory):

cd                                    # make sure you're in your home dir
ln -s /scratch/dduck1/largefile .     # link largefile back to current dir (.)

See also:

Back to SwatCS Help Docs