this paper, we analyze, through simulation, the potential benefits of having the server also parse the document and pre-fetch into its main memory cache the inlines that will be requested by the remote client. The parameters for the simulation were obtained from a detailed workload characterization of a Web server available to the authors. Response time improvements of up to 48% can be achieved with even moderate cache sizes. 1. Introduction The Web has experienced an exponential growth since its inception. Recent statistics from a 1997 Nielsen Media Research study put the number of Web users in the US at 45 million and in Canada at 5 million. Applications such as electronic commerce and digital libraries will help to further increase the growth rate. Electronic commerce is forecast to be a $6.9 billion industry by the year 2,000. Currently, 85% of America's Fortune 100 companies have a presence on the Web and spend between $0.75 and $1.25 million on their Web si

In order to meet the quality of service demanded by a growing number of online customers, e-commerce services need to use scalability techniques. This paper introduces the concept of e-commerce representatives, a means of scaling the performance of e-commerce services. Erepresentatives are programs that execute on a cache server or at nearby machines. E-representatives can be implemented using redirection, a mechanism available in popular cache servers. Using analytical and simulation models, we show the potential performance gains obtained by e-commerce sites that distribute their services among erepresentatives and contribute to reduce bandwidth consumption and network latency. Keywords: e-Commerce, QoS, scalability, simulation, analytic models 1

Abstract. Latency is a fundamental problem for all distributed systems including digital libraries. To reduce user perceived delays both caching – keeping accessed objects for future use – and prefetching – transferring objects ahead of access time – can be used. In a previous paper we have reported that caching is not worthwhile for digital libraries due to low re-access frequencies. In this paper we evaluate our previous findings that prefetching can be used instead. To do this we have set up an experimental prefetching proxy which is able to retrieve documents from remote fulltext archives before the user demands them. Using a simple prediction to keep the overhead of unnecessarily transfered data limited, we find that it is possible to cut the user perceived average delay a factor of two. 1

The web has experienced phenomenal growth and has become an important element of the information infrastructure of many organizations. Web performance and around-the-clock availability are the major concerns of Webmasters of sites that support mission critical applications, such as electronic commerce. Web sites are complex computer systems consisting of many interconnected computers, networks, and routers. Some of the computers have specific functions such as certification or caching. Workloads on Web sites are often unpredictable in nature. Heavy load spikes and a large variability of the objects requested over the Web are unique characteristics of Web workloads. This paper reviews the important issues and challenges in modeling Web-based systems. Workload characterization, predictive models, and their use in various situations are discussed.