Enhancements To The SAM-grid Infrastructure
Abstract
SAM-Grid is a grid computing infrastructure for high energy physics (HEP) experiments in Fermilab. It is composed of data handling, job management and information management components. There are several challenges when the number of sites participating in the experiment increases. This thesis presents the enhancements made on the SAM-Grid infrastructure. This includes scalability and performance related enhancements.
The enhancements mainly affect the batch adapter, monitoring and security layers. As a scalability aspect, Monitoring and Information services required changes to make it easier for monitoring large number of jobs. SAM-Grid is integrated with Sun Grid Engine (SGE) batch system. The batch adapter layer corresponding to SGE is different from that of other batch systems like PBS. SGE is successfully being used for production in one of the execution sites. Integration of MyProxy with SAM-Grid is one of the performance related enhancements. SAM-Grid system performance will improve because job failures due to proxy expiration are avoided.