Elms backup strategy

This page describes the backup strategy for server and workstation data at the Elms

The two 'always on' servers (sx and anubis) both have software raid arrays with a small OS and Swap partition and a single large data partition. The data can be characterised as:

The data can be further categorised as

Since backup space is limited and there is some redundancy in the raid array for the common scenario of hard disc failure, no attempt will be made to backup the continence data beyond storing it on a raid array.

Critical data backup strategy - Small files rarely changing

These files will be copied to the backup server (anubis) using a robocopy script set to mirror directories. The script will be run weekly from the backup server. The backup server will have VSS installed and configured to take a weekly snap shot at a time which will not conflict with the backup. VSS gives the ability to 'go back in time' for over a year (64 weeks) space permitting.

Backups will be placed in the following directory structure

Backup
  Machine1
    Dir1
    Dir2
    ...
  Machine2
    Dir1
    Dir2
    ..

Critical data backup strategy - Large files rapidly changing

The above strategy would not work well for these files because they would cause the VSS history buffers to overflow quickly, severely reducing available backup history. The same strategy will be used, however, a separate volume will be set aside (on a workstation) for these files.

Scheduled activities

06:00 - 01:00 Anubis records Radio 4
01:05 Anubis fetches schedules
01:10 Sx installs patches and reboot
02:00 Virtual machines install patches and reboot
03:00 Sunday weekly backup scripts run