Caching Dynamic Content Rajiv Motwani and Soila Pertet Electrical & Computer Engineering Department Carnegie Mellon University 5000 Forbes Avenue, Pittsburgh, PA 15213-3890. {rmotwani, spertet}@andrew.cmu.edu ABSTRACT The generation of dynamic content on a web server can contribute for a significant portion of the total service time for client requests. Caching dynamic content could yield a significant improvement in average response times with the added advantage of reduced load on the server. In this paper we survey different dynamic content-caching strategies and investigate their effect on throughput and client response-times. We focus on strategies that can be deployed at the web server or reverse proxies controlled by the content-provider. For instance, we evaluate the effects of cache size, cache replacement strategies such as LRU (LeastRecently-Used) and FIFO (First-In-First-Out) on server load and latency. We discuss the effects of caching granularity on server load, latency and cache efficiency. We also discuss the trade-offs associated with different caching strategies and address the issue of "when" and "where" to cache dynamic content.