A New Approach of Prediction of Memory Leak in the Cluster Computing Applications

This paper presents a memory leak prediction algorithm for the cluster computing applications. This proposed algorithm uses process characteristics to calculate the exact memory requirement and uniquely identifies maximum memory utilization of an application before the application starts its execution. During the appffplication execution phase, memory leaks in the application processes in the cluster is identified by existing Dynamic Memory Monitoring Agent (DMMA) gives information to the end users to make corrective actions and removes memory leak processes from the affected nodes. This unified approach increases the reliability and fault tolerant in the cluster computing.