NSMS logo

Case Study: Improving emergency status notification

In August 2007 the electrical power to the OUCS machine room had to be suspended for a significant part of a day for a routine five yearly safety inspection. It was impossible to supply sufficient power from alternative sources to keep all the services going but there were some essential services that had to be switched over to an auxiliary supply. The Status service (http://status.ox.ac.uk) provided by OUCS was one of these essential services.

The Status server is a Debian Linux based system normally running as a VM on the NSMS blade hardware but this hardware was going to be powered down during the shutdown period. The following plan was devised to provide continuity for the Status service:
  • A small server, able to be powered from the auxiliary supply, was set aside to support the temporary service.
  • Windows 2003 was already installed on this server and VMware Server (free software) was installed on this.
  • The live VM that was running on the blade system was copied to a file on the temporary server.
  • Although the production blade system runs VMware ESX, the VM files for that system are compatible with the VMware Server program so the VM file of the status system was immediately able to run on the temporary server.
  • Before the main power down, the blade system was powered down. As soon as the Status Service VM was powered down, the copy installed on the temporary server was started. The temporary server was connected to the same network domain as the blade system so no DNS changes were required; the status service was immediately available.
  • After the safety inspection was completed, the temporary Status service VM was powered down and the service was restarted on the blade system.

Up: Contents Previous: Case Study: Enhancing web services at the Museum of the History of Science Next: Case Study: Building An Enterprise Infrastructure Without An Enterprise Budget