Skip to Main Content
Object replacement strategies are considered as key performance factors of a web proxy caching application. Several algorithms have been proposed, analyzed and simulated during the last decade, in order to decide upon the champion one. In this paper, we recommend the methodology of Data Envelopment Analysis as a powerful tool, with a complementary activity to that of trace driven simulation, which is able to discover the set of the `good-enough' replacement strategies - i.e. strategies that exhibit good results for the basic performance metrics - under a variety of simulated workloads.