Data Backup Policy
Views:
Contents |
Policy Statement
All data within the IS systems shall be protected against loss or corruption through the use of robust backup procedures. These procedures will be performed sufficiently frequently to ensure that the organisation is not significantly disrupted should a failure occur within the systems.
Full backups should be performed by members of the IS department on a daily basis for all operational systems. Backup procedures should have built-in verification and be tested to ensure that the stored files can be restored if needed.
Apart from a few exceptions, no data files are kept in workstations, removing the need to back them up individually. Backups are only performed at the server level. When files on a workstation become corrupted, the workstation is completely rebuilt, and all information is lost. This also happens on a regular basis in order to distribute software upgrades. Because all backups occur on servers, it is the responsibility of the members of the IS department that they take place following the procedures detailed below.
A distinction has to be made between files that are essential for the functioning of the information systems and the college's flow of information, and user's personal files. Ultimate responsibility for data stored in a user's home directory lies with that user. Whenever possible, users should keep copies of their own personal files.
Backup Procedure
Symantec backupexec software for Netware, Linux, Mac and Windows
The following backups are taken:
- full, Friday backup of servers ATHENA, CRIUS, APOLLO and ATLAS, THAY, SKARO, TETHYS, DAVROS, ARES, PHOBOS. On all other days an incrimental backup takes place.
- BROMCOM backs up the database every night. The data backed up is listed in the selection list for each job.
- partial regular backups via a backup agent of Gorgon to Sobek (backu Server)
SQL Databases on ATLAS are backed up every day, in the case of Resource 32000 (MIS) this is backed up 2 hourly.
The backup server backs up all the information to a tape library on a Saturday night.
Any anomalies and failed jobs should be reported and remidied as soon as possible
Backup Tape Labels
Backup tapes are labeled in the following format:
[SERVER]-[DAY_OF_THE_WEEK]
The tapes should be formatted accordingly with the name that appears on the label.
There are five full backups per week and per server. There are eight tapes per server:
- [SERVER]-MONDAY
- [SERVER]-TUESDAY
- [SERVER]-WEDNESDAY
- [SERVER]-THURSDAY
- [SERVER]-FRIDAY1
- [SERVER]-FRIDAY2
- [SERVER]-FRIDAY3
- [SERVER]-FRIDAY4
Monday - Thursaday Backup Procedure
Incrimental backups are performed Monday through to Thursday at 22:00.
Check the logs for the previous night’s jobs on all servers and report/remedy any anomalies.
Friday Full Backup Procedure
Full backups are performed for each server each Friday at 22:00. There are four Friday tapes that provide backups going back for a month. The order in which they are used is set up in a rota:
Backup Rota
First thing in the morning on Friday, do the following:
Check the rota to determine the tapes needed for all of the weekend full backup jobs
Check the logs for the previous night’s jobs on all servers and report / remedy any anomalies.
Then, on Monday morning:
Remove the tapes from the previous Friday from the drives. Tapes will be taken home until they are needed again (see rota).
A whole set of Weekend tapes for the previous week are stored offsite until a new set is available. This is a requirement of our Disaster Recovery policy.
Cleaning the Tape Drive
If there is an orange light flashing on the tape drive, this means that the drive needs cleaning. Just put a cleaning tape in drive and wait until it runs (it will do so on its own). This takes about one minute. Mark off one square in grid on cleaning tape to ensure tape is only used for recommended number of times.
Backup From Workstations
In certain cases, it might be necessary to backup information from certain workstations running specific software (LSC applications…). The relevant files are to be copied to an area on a server, and included in the daily backup.
Restore Procedure
At least once a month a restore job should be run to restore parts of the backups from the tape to a set directory on the servers. This is to be done on all servers, and its success/failure to be recorded on a log file.
Timescale for restoring critical files is specified in the Disaster Recovery Policy
