Job Monitoring in an Interactive Grid Analysis Environment

Abstract

a complex system like the Grid, job monitoring is essential for understanding its operation, debugging, failure detection and for performance optimization. To achieve this, data about the jobs that are currently executing in a Grid must be gathered and processed to reveal important information about the job progress. Then, according to the results, the system may need to be controlled. It is the task of the monitoring system to provide facilities for this. The monitoring system must be able to provide information about the current state of jobs, such as status of job, priority, amount of IO performed by the job, node at which job is scheduled etc, as well as to provide notifications when certain events (e.g. job completion or termination) occur. Interactive grid environment has emerged as a next generation Grid environment which will enable users to harness heterogeneous collection of computer components while giving them full control over their jobs. Users in this environment will be able to monitor their jobs while they are executing and direct the execution of jobs to different nodes based on the job monitoring information. But in order to fulfill these requirements certain applications and services are needed that can support job monitoring and job steering. So a Job Monitoring Service has been developed that monitors the jobs executing in a Interactive Grid Environment. Job Monitoring Service will provide an easy to use API through which a client can obtain monitoring information about a job. The purpose of Job Monitoring service is to provide real time job monitoring and status feedback to other services in Interactive Grid Analysis Environment while operating in close interaction with execution service to provide interactivity, fault tolerance and traceability in an Interactive Grid Analysis Environment.