Skip to Main Content
In present computing plethora, the cluster technology seems to gain popularity day by day. The prime reason for this popularity is high availability, increased reliability and high performance through efficient resource usage. To realize these benefits, the processing capacity of various nodes in a cluster must be allocated fairly. Often the clusters suffer from under utilization due to inappropriate choice of scheduling policy. The goal of the scheduling is to exploit the true potential of the system. Hence, finding the appropriate granularity of tasks and distributing them in such a way so that each machine is assigned equal work and thus, balancing the load across the cluster, is major issue of concern. Usually, it is seen that local processes get priority over remote processes and migrated (remote) processes begin to starve. We propose an optimized scheduling approach for migrated processes that ensures reduced latency time along with no starvation policy for any (local or remote) process. Our experimental results have been able to establish that priority and criticality based scheduling approach can increase overall system throughput by about 30-50 percent.