Generating Representative Web Workloads for Network and Server Performance Evaluation-英文文献.pdf

Generating Representative Web Workloads for Network and Server Performance Evaluation Paul Barford and Mark Crovella Computer Science Department Boston University 111 Cummington St, Boston, MA 02215 fbarford,crovella g@ BU-CS-97-006 REVISED Decemb er 31, 1997 Abstract One role for workload generation is as a means for understanding how servers and networks resp ond to variation in load. This enables management and capacity planning based on current and pro jected usage. This pap er applies a numb er of observations of Web server usage to create a realistic Web workload generation to ol which mimics a set of real users accessing a server. The to ol, called Surge (Scalable URL Reference Generator) generates references matching empirical measurements of 1) server le size distribution; 2) request size distribution; 3) relative le p opularity; 4) emb edded le references; 5) temp oral lo cality of reference; and 6) idle p erio ds of individual users. This pap er reviews the essential elements required in the generation of a representative Web workload. It also addresses the technical challenges to satisfying this large set of simultaneous constraints on the prop erties of the reference stream, the solutions we adopted, and their asso ciated accuracy. Finally, we present evidence that Surge exercises servers in a manner signicantly dierent from other Web serve