Cluster configurations are a cost effective ways which have become common options to enhance several classes of applications in many organizations. However, the load balancing approaches to these clusters suffer from significant deficiencies: Some load balancing algorithms does not consider the communication delay involved in collecting load information, while others depend on cluster size that makes them less scalable.To overcome these deficiencies, an efficient load balancing technique named as "Recent Neighbor" (RN) has been presented. In this article, it is assumed that the physical servers of the cluster are heterogeneous with respect to processing power and workload.The simulation results show that the "Recent Neighbor" (RN) load balancing algorithm yields better performance when compared with the so-called "Nearest Neighbor" (NN) load balancing algorithm. The project has been implemented in. Net environment using C# language. The "Recent Neighbor" load balancing algorithm is dynamic, de-centralized and symmetric-initiated.
展开▼
