WACA: Workload and cache aware load balancing policy for web services

Abstract

The popularity of the web applications has shown rapid growth on Web Services and Cloud Computing. One of the major challenge in Web applications is that it need to process massive files and data-sets to respond to an increasing number of users requests. These data may range from multimedia data to sensor data. Along with the size of data, user satisfaction is an important constraint imposing Quality of Service requirements. Therefore the use of web clusters becomes necessary for numerous applications, fostering the need of efficient load distribution strategy. This article presents a new load balancing policy called Workload And Cache Aware load balancing (WACA) policy aimed at reducing the information retrieval time of the processor using Bloom Filter Based cache summaries. The WACA is a centralized load balancing algorithm which routes the incoming requests to the machine where the underlying data will most probably be in its cache without overloading any nodes in the cluster. The proposed policy was developed and tested on a load balancing framework called CLOUDIZER, and the performance of the strategy was compared over the traditional algorithms.