Grid aware HA-OSCAR

Physicists today have employed grid technology to overcome various resource level hurdles. The collective resource utilization achieved through grid computing is critical to the overall computing capacity of the community and should be guaranteed. In an environment where job sites are cluster systems, a service node failure renders a whole system outage. Our grid-aware HA-OSCAR effort was motivated by the popularity of the cluster architecture in the grid environment. We propose the high-availability architecture, HA-OSCAR, for cluster-based job sites in the grid environment. This architecture deals with fault tolerance at the service level complementing task-based solutions such as checkpoint/restart. We discuss various service availability issues related to the grid, some issues and preliminary results obtained while implementing the smart failover feature and the automated grid installation package. Our report entails the performance benefits achieved after applying the HA-OSCAR solution to the cluster components of the grid compared to regular Beowulf style cluster solutions.