Workload characterization is a critical component in capacity planning for Web-based n-tier systems. Web workloads are characterized by different patterns and intensities of different transactions. In this paper, we present a technique for characterizing Web workloads in a transaction-oriented manner through offline analysis of Web server logs. The results can be used in performance evaluation of the current system, extrapolating the load for increased user base for capacity projection of an augmented system, and in preparing performance testing strategies for the augmented system.
展开▼