Abstract

Proxy caching has been extensively used to cache web contents and has achieved greatly reduced response latency, shared web server load, and minimized network traffic. Recently there has been a major shift in the usage of the Web. More and more web pages are generated dynamically by web processing components. These web pages are non-cacheable. Thus, no matter how good the existing caching scheme is, the cache hit ratio is limited due to the increasing percentage of non-cacheable objects.

In this dissertation, we develop a weblet environment to facilitate proxy caching of web processing components. In the weblet environment, the processing components at web servers are implemented as weblets. These weblets can migrate from web servers to proxy servers to perform required computation and provide faster responses. A weblet migration protocol, weblet caching strategies, a uniform weblet execution platform, a weblet based request processing protocol, and security protection mechanisms have been developed to enable proxy servers to support weblet migration and execution.

Most of web applications need to access databases. To maximize the benefits of the weblet environment, data frequently accessed by weblets should also be cached at proxy servers. Since a proxy server usually has limited disk space, the size constraint of the proxy server is considered to select data in a database. A quantitative analysis approach has been developed to guide the cache table selection process. To achieve more effective data caching for cache tables, a data selection algorithm has been proposed to partition cache tables and select the highly accessed partitions to be cached.

Our research has a significant impact on the web system infrastructure. With the current trend, we believe that more and more web applications will be computation oriented. With existing proxy protocols, the cache hit ratio will drop due to these processing components. With the weblet approach, these processing components can be migrated and cached and a higher proxy cache hit ratio can be achieved. Thus it can achieve significant performance improvement. Also, the weblet concept complements current web caching techniques and can be easily integrated with existing web performance improvement approaches.