Building a Scalable Web Server with Global Object Space Support on Heterogeneous Clusters

• Authors: Ge ChenCho-li WangFrancis C. M. Lau
• Venue: CLUSTER '01 Proceedings of the 3rd IEEE International Conference on Cluster Computing
• Year: 2001

Clustering provides a viable approach to building a scalable Web server system. Many existing cluster-based Web servers, however, do not fully utilize the underlying features of the cluster environment, and most parallel web servers are designed for homogeneous clusters. In this paper, we present a pure-Java-implemented parallel Web server that can run on heterogeneous clusters. The core of the proposed system is an application-level "global object space", which is an integration of the available physical memory of the cluster nodes for storing frequently requested objects. The global object space provides a unified view of cluster-wide memory resources, and allows transparent accesses to cached objects. Using a technique known as cooperative caching, a requested Web object can be fetched from a node's local memory cache or a peer node's memory cache to avoid hot spots and excessive disk operations. A preliminary prototype system has been implemented by modifying the W3C's Jigsaw Web server. We obtained good speedups in the benchmark tests, indicating that clustering with cooperative caching can greatly improve the performance of a Web server system.