An efficient resource optimization in intra-tenant heterogeneous hadoop cluster

Apache Hadoop is an open source big data processing framework. Hadoop helps to Store and process big data. YARN efficiently share resources among the nodes in cluster. One of the key building blocks of any heterogeneous cluster computing is resource fairness. Several researchers propose models to improve efficiency of the resource utilization by sharing resources fairly among the nodes. Memory less Resource Fairness (MLRF) widely used in existing frameworks like YARN, Mesos and Dryad. Long Term Resource Fairness (LTRF) improves Resources Fairness but not up to the mark. This paper propose Combined Regression Ranking (CRRYARN). It predict ranking of different workloads based on amount resource utilization factors like CPU Intensive, Memory intensive etc., Based on this ranking the resource sharing happened among the workload in Single Tenants of the same cloud. It works on loan lending with interest free, rank based discount share model. It improves performance of the existing model by 30% effectively.