IT Development/Proposals/Backup

From Wikimedia UK
Jump to navigation Jump to search
IT Development
Main pageInfrastructureDocumentation / ToolsPortfolioTechnology CommitteeProject requests


This proposal outlines Wikimedia UK's requirements for a backup and disaster recovery solution, encompassing the web/email servers and office computers.

nb this is currently a draft


Infrastructure[edit | edit source]

The WMUK infrastructure consists of two main sections:

  • Three virtual servers, hosted with Rackspace
  • Personal desktops, volunteer laptops and one server, located in the London office

The virtual servers are located on a private LAN within Rackspaces' London data centre and provide essential services such as web and email. The office machines are on a local network in the London office, administered by XXXXXX. Wikimedia UK computers have a variety of operating systems, including Windows, OSX and Linux.

Requirements[edit | edit source]

An overall solution is required to securely backup essential Wikimedia UK data for the purposes of security and disaster recovery. The key requirements are:

  • Secure backups: data must be securely encrypted
  • Offline backups: ability to place recent copies of backups in a fire safe
  • Easy to administer: office staff should be able to conduct offline backup procedure, and the system should otherwise require little to no maintenance
  • Open Source: in line with Wikimedia UK's commitment to using Open Source software when appropriate
  • Comprehensive: capable of backing up server information as well as local files
  • Disaster Recovery: the system should form part of an automated disaster recovery plan
  • Cross Platform: the system should be capable of backing up Windows/Linux/Mac
  • Multiple backup oprtions: capability to handle files, as well as databases

Proposed system[edit | edit source]

The proposed backup scenario requires backups to be immediately available for online recovery of servers, offline (in the office) for backup safety, and "offsite" (in a fire safe) to additional security. We can term these formats as short, mid and long term backups.

Infrastructure requirements:

  • Additional Rackspace cloud server to host backups
  • Office server for backups
  • Fire safe & disks to store offsite backups

The current office server is probably sufficient for purpose, although consideration will have to be made as to whether it is advisable to have file shares and backups on the same physical machine.

Logistics:

  • Rackspace servers will be backed up daily (the most regular changes will be MySQL, and file uploads) to the cloud backup server
  • This will be synced overnight to the office server
  • Once a week (??) office staff will archive backups from the office server

Specialist backup software would need to be installed on the cloud servers. Likely all that is needed locally is to have a Cron job synchronize the cloud data locally (using e.g. rsync), although this depends on the in-office backup requirements. A user friendly tool could be used to allow staff to archive backups.

One key aspect will be email notifications of successful backups, and any errors.

Solutions[edit | edit source]

There are various open source solutions available to meet Wikimedia UK's server backup needs:

Recommendation[edit | edit source]

Todo[edit | edit source]

  • Clean up config files and get them version controlled
  • Set up something like Puppet or custom shell scripts for disaster recovery
  • Identify files/data for backup