Automation in Validation of HPC Scheduler

Abstract

Dynamic Grid Scheduler of Middleware - in HPC (High Performance of computing) has to perform the very critical task of Scheduling a very huge amount of computeintensive jobs across multiple data centers that are available across the Globe. It is very important to make sure that all jobs are going to land as per specified requirements. Manual validation of this whole scheduling process becomes more tedious and time-consuming. By the Automation validation scripts, we can validate the same process in a shorter time with more accuracy. Apart from that to provide efficiency in a huge amount of job submission to the main scheduler, one flow manager is there who will take care of the submission of bulk jobs. The flow manager provides the facility to monitor the execution of those bulk jobs also. Flow manager is supporting a different job block, those jobs can be directly submitted on the executing machines. By this utility, we can reduce the total execution time of bulk jobs. As Flow-Manager needs to submit jobs to Scheduler, it is highly expected that the Flow-Manager performs well even with high load. To ensure that the performance of the ow-manager also needs to be validated.