Maintaining replicated redirection services in Web-based information systems

Abstract

The growth in the commercial use of the Internet and the proliferation of new Web information services has led to a number of problems - overloaded servers, traffic congestion and increasing client latencies. Dynamic replication of information among geographically distributed Web servers is commonly used to address such problems of scale. A key challenge for such systems is to maintain an efficient and scalable redirection service that tracks the servers where documents are replicated and assigns each client request to a server that can offer the "best" service. We propose a new scheme for maintaining redirection services in Web-based information systems. The proposed scheme is based on adaptive replication of the entries of the location directory that provides the redirection service. Network-level support enables client requests to be redirected to appropriate servers in a quick and efficient manner. We integrate our scheme with a dynamic object replication protocol proposed in the literature. Simulations based on synthetic workloads and the Internet backbone map of CRL Network Services demonstrate that the replicated redirection service delivers significant improvements in terms of client response time, while successfully eliminating hot spots in the network and imposing minimal traffic overhead.