Skip to content

RDS Storage

diagram

RDS implements two, independent, disk-based copies of all stored data, providing very high levels of data resiliency and availability. A third, tape-based, disaster-recovery copy is taken daily, as shown below. The two disk copies are physically separated, in different halls and for every file written, copies are stored, almost simultaneously, in both locations.

Backup

Snapshots

RDS implements snapshotting of all data, a lightweight mechanism whereby read-only copies of files are taken at defined intervals. The standard snapshot scheme for a Project is

snapshotting interval

  • Hourly, for 1 day
  • Daily, for 1 week
  • Weekly, for 1 month
  • Monthly, for 1 year

Snapshots provides a level of protection from accidental file deletion/corruption and ransomware attacks. Snapshots are accessible to users read-only. A ransomware attack on a researcher’s machine that is found to have encrypted the data can therefore be foiled, for the most part, by restoring from the last snapshot taken before the encryption started.

Offline backup

A daily copy of all RDS Projects is made on a separate Disaster Recovery (DR) cluster. The DR cluster copies disk changes to a tape library providing the security of physical separation and a different media format.