Journal of Computing
This paper proposes distributed shared memory cluster architecture with load balancing. The architecture is based on dynamic task scheduling approach for distribution and assignment. It enhances the performance of communication across clusters for data access. The proposed dynamic load balancing model uses the concept of work stealing, which intelligently balances the load among different nodes. The paper stealing consistently provides higher system utilization when many jobs are running with varying characteristics. This results in efficient use of the system. The performance analysis shows the proposed architecture to outperform the previously proposed distributed shared memory clusters in terms of scalability and efficiency.