Backup Plan

Backup Plan

1. Policy Statement

As part of McMaster University Library (MUL) implementation of the Bit-stream Copying Preservation Strategy (as detailed in the Preservation Implementation Plan), MUL is committed to regular backup procedures of both data storage areas and its operational areas (e.g. databases, application files). These backups are intended to serve as the basis for restoration of the Library’s digital repositories (Digital Archive and Closed Archive) materials in the case of disaster recovery or corruption of data.

Data backup at MUL is coordinated by the Library Digital Experience Unit. The data is stored on physical hardware located in the MUL data centre.

2. Implementation

2.1 Database Backup

This backup strategy applies to content that is stored in a database. Primarily, this refers to objects located in the Library’s digital repositories’ (Digital Archive and Closed Archive) databases (MySQL).

2.1.1 MySQL Database Backup Strategy

  • Database backups happen daily and are saved to a sub-folder of the Fedora installation folder. They are then backed up alongside the data folder as described below under “Fedora Commons Backup Strategy”

2.2 Virtual Machine (VM) Backup

  • The virtual machine on which the Digital Archive runs, is not currently backed up. This VM is pending migration to a new host, on which a snapshot of the VM will be backed up weekly.

2.2.1 Fedora Commons Backup Strategy

  • Fedora Commons is backed up incrementally once per week to an on-site storage array. Backups are currently retained indefinitely.

2.2.2 Drupal (Islandora)

  • Drupal is backed up incrementally once per week to an on-site storage array. Backups are currently retained indefinitely.

2.3 Objects

  • Fedora objects are backed up incrementally on a daily schedule to an on-site storage array. The local backup is synced to cloud storage, coordinated by Ontario Council of University Libraries (OCUL), once per week. Backups are currently retained indefinitely.

2.4 Verification

  • Backups are stored to ZFS volumes, which ensure data correctness and protect against bit rot by storing a checksum for each file. A scheduled monthly process known as “scrubbing” reads each file on the volume and verifies its checksum. If a file is found to have become corrupted, error-correction is automatically performed.
  • There is currently no other verification of backups or testing of restore procedures performed. MUL’s IT staff have recognized this deficiency as a priority area of improvement for 2020.


January 07, 2020

Digital Archives Advisory Board, McMaster University Library