Profile-based data delivery for web applications

Abstract

An increasing number of clients use the web for applications such as webbrowsing, E-commerce, and webportals. An important characteristic of data access in this environment is that clients often have different requirements for the latency and recency of objects. However, there has been little work that aims to accomodate varying client needs. This Ph.D. research introduces client profiles that allow clients to express their requirements with respect to the latency and recency of objects. Profiles are used by a cache or web portal to determine when to download a fresh object and when to use a cached copy. Profiles can be tuned to reflect the requirements of a client or application. The profile-based framework is straightforward to implement for both clients and caches, and can scale to a large number of clients and applications. Preliminary results show that profiles can reduce latency and bandwidth consumption compared to existing techniques, while still providing recent data in many cases. © Springer-Verlag Berlin Heidelberg 2002.