ADAPTIVE LOAD BALANCING FOR CLUSTER ARCHITECTURE USING TRAFFIC MONITORING WITH CONTENT AWARENESS

Abstract

With the rapid growth of both information and users on the Internet, how to effectively improve the quality of network service becomes an urgent problem to be addressed. Load balancing is a solution to this problem in an effective way. Different adaptive load balancing methods have been developed to estimate servers load performance. However, they suffer from either increased processing load on the servers, or additional traffic on the network and the servers, or are impractical in real time scenario. Need of high scalability, high reliability and high availability are key issues in load balancing. Load balancing algorithm along with cluster architecture fulfills all the needs. In this dissertation, we introduce the concept of content based queues, and have used RTT passive measurement technique to get an adaptive load balancing algorithm inside the cluster. Using the same queue for each type of request results into overload on server, so we introduce the concept of different queue for different type of request. The whole load balancing task is performed by a webproxy inside and outside the cluster, a proxy that has access to all servers so it also removes the drawback of dynamic algorithm in which most of the load balancing task is performed by server itself.