Server Operations: Backup Policies

One of the Server Operations tasks focuses on the production servers, which are either backed up via TSM or AFS. The primary purpose for all backups is to recover from hardware failures and other systemic events. In cases of data accidentally lost by a user, Server Operations will make a reasonable effort to help recover the data.

TSM Backups

The production servers are backed up once a day. However, there are a number of reasons for a server's backup to fail on any given night.

Backups are retained for eleven weeks. Shortly after that they are automatically deleted.

Decommission a Server
In addition to the regular backups, Server Operations makes a "final" backup before decommissioning a server. This backup is kept for at least a year and will be manually deleted sometime after the year is up.

AFS Backups

The following policies pertain to the production AFS cells (athena and dev) maintained by Server Operations. Other test cells may have fewer backups (e.g., ops cell), or no backups at all (e.g., zone cell). Additionally, the following policies do not apply to volumes or lockers that are designated as non-backed-up.

Frequency for Regular Backups

  • There is at least one backup from every two-week interval for the last four months. Server Operations makes backups more frequently than this, but some are only kept for a short period of time.
  • When a backup is no longer needed, either because it is older than four months, or because there are enough other backups from the same time period, Server Operations will recycle or destroy the tape or hold it for future re-use or destruction at their discretion.

Archival Backup

  • In addition to the regular backups, Server Operations makes an "archival" backup twice a year, shortly after the end of each term.
  • Archival backups may include volumes and lockers that are designated as non-backed-up.
  • These backups will be kept for a minimum of five years. Like the regular backups, once "archival" backups are old enough, they may be recycled, destroyed or held.
  • If at any point in the life of a backup Server Operations no longer has the necessary tools to read a backup, it will be immediately eligible for re-use or destruction, regardless of the above policies.