Efficient Algorithms for 1- and 2-echelon Web Proxy Location Problems

With the increasing popularity of the World Wide Web (Web), it is very difficult for a single popular Web server to handle the explosive demand from its clients. Web caching (proxy) is an effective technique to reduce a Web server’s load, alleviate Internet traffic, and reduce user’s response time when accessing a Web service. A fundamental question is where to locate proxies throughout the network to minimize total access time. We consider the problem of locating k Web proxies on a network of n nodes. The objective is to minimize the overall cost for accessing Web services to satisfy a given level of demand. The location problem is known to be NP-hard for general network topologies. We present O(n) time algorithms which computes optimal proxy locations for the ring topology for single-echelon location problem. We present an O(k l kgn ) time algorithm for the linear topology for the 2-echelon model, where kl and kg, denoting the desired number of local and global proxies, respectively. We develop heuristic algorithms for finding proxy locations for a general network topology and compare the heuristics with optimal solution through simulations.