Revisiting Web Server Workload Invariants in the Context of Scientific Web Sites

Abstract

The Web has evolved much from when Arlitt and Williamson proposed the ten Web workload invariants more than a decade ago. Many diverse communities now depend on the Web in their day-to-day lives. A current knowledge of the invariants for the Web is useful for performance enhancement and for synthetic Web workload generation. Invariants can also serve as a useful tool for detecting anomaly and misuse, a new dimension of Web usage arising from the change in trust assumptions in the Internet in the recent years. Focusing on scientific Web servers, we revisit the Web server workload invariants and find that only three out of the ten invariants hold as-is. We investigate appropriate revisions to the invariants that do not hold and also propose three new invariants for scientific Web servers