Overview
The departmental backup system is designed to regularly back up data (not software and operating systems) to a disk & tape system hosted over the road in the Gurdon Institute. For the purposes of these pages "Backup" (as opposed to "archival") is copying data from rapid access systems onto cheaper storage in order to recover from accidents or system failures. Copies are only kept for a limited period of time. Systems to be backed up must be on the wired network and be turned on during the backup period.
Server backup jobs are started in the evening and desktop backup jobs are started at lunchtime. A given dataset must not be actively being copied for a new job to be allowed to start. The priority is:
- Quarterly, based on a queue, a Full copy is taken of each dataset. For gigabit network connected servers 1 terabyte takes about 5 hours to copy offsite, verify and stored to offsite tape. During the pandemic old copies are being moved onto onsite tape as offsite tape becomes full, we have limited access to the offsite location.
- Fortnightly a Differential copy is taken of each dataset. This copies all files which have changed since the last Full copy. This is initially stored on offsite disk, but older jobs are moved onto onsite tape, keeping at least one offsite copy at all times. Copies are kept for at least 3 months.
- Daily, when no other backup jobs are using a dataset, an Incremental copy is taken. This only includes files which have changed since the next most recent backup (of any type.) This is stored on offsite disk for at least one month.
I get a report on every backup that occurs and check a random file every month to ensure things are ticking over OK. Designated lab representatives can also see what is backed up, when it was backed up and where it is.
Data can be excluded from backups by creating a file ".nobackup" in a folder which will exclude all the contents of the folder.
Do you need it?
- If you're storing data on larch then your data is already backed up except for contents of folders containing a ".nobackup" file or folder.
- If you are using a PC with the "DEPT" login, data in your "Documents" and "Desktop" folders is on the server "Larch" so is already backed up (See #1 above.) If you are storing elsewhere then it will not be.
- Some lab servers are backed up too. Ask in your group. If no one knows then you have a problem in your lab.
Compared to using a hard disk
- Data offsite can only be written to by the backup software. If you get a virus that attempts to destroy or encrypt your data it can't directly attack backups through this system
- If you are backing up to a USB disk or NAS viruses can quite quickly delete and overwrite things.
- It is slower and more complicated to recover data from the backup system.
- Doing both is not harmful
- If you manage your own backups you need to check they are working.