Generating representative Web workloads for network and server performance evaluation

ABSTRACT

One role for workload generation is as a means for understanding how servers and networks respond to variation in load. This enables management and capacity planning based on current and projected usage. This paper applies a number of observations of Web server usage to create a realistic Web workload generation tool which mimics a set of real users accessing a server. The tool, called Surge (Scalable URL Reference Generator) generates references matching empirical measurements of 1) server file size distribution; 2) request size distribution; 3) relative file popularity; 4) embedded file references; 5) temporal locality of reference; and 6) idle periods of individual users. This paper 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 properties of the reference stream, the solutions we adopted, and their associated accuracy. Finally, we present evidence that Surge exercises servers in a manner significantly different from other Web server benchmarks.

REFERENCES

• 1: Virgílio Almeida , Azer Bestavros , Mark Crovella , Adriana de Oliveira, Characterizing reference locality in the WWW, Proceedings of the fourth international conference on on Parallel and distributed information systems, p.92-107, December 18-20, 1996, Miami Beach, Florida, United States
• 2: Martin F. Arlitt , Carey L. Williamson, Web server workload characterization: the search for invariants, Proceedings of the 1996 ACM SIGMETRICS international conference on Measurement and modeling of computer systems, p.126-137, May 23-26, 1996, Philadelphia, Pennsylvania, United States
• 3: Henry Braun. A simple metllod for testing goodness of fit in tile presence of nuisance parameters. Journal of the Royal Statistical Society, }980.
• 4: Tim Bray, Measuring the Web, Proceedings of the fifth international World Wide Web conference on Computer networks and ISDN systems, p.993-1005, May 1996, Paris, France
• 5: The Standard Performance Evaluation Corporation. Specweb96, http://www.specbench.org/org/web96/.
• 6: Mark E. Crovella , Azer Bestavros, Self-similarity in World Wide Web traffic: evidence and possible causes, Proceedings of the 1996 ACM SIGMETRICS international conference on Measurement and modeling of computer systems, p.160-169, May 23-26, 1996, Philadelphia, Pennsylvania, United States
• 7: Carlos Cunha , Azer Bestavros , Mark Crovella, Characteristics of WWW Client-based Traces, Boston University, Boston, MA, 1995
• 8: Ralph B D'Agostino , Michael A Stephens, Goodness-of-fit techniques, Marcel Dekker, Inc., New York, NY, 1986
• 9: S. Deng. Empirical model of WWW document arrivals at access link. In Proceedings of the 1996 IEEE International Conference on Communication, June 1996.
• 10: Ashok Erramilli , Onuttom Narayan , Walter Willinger, Experimental queueing analysis with long-range dependent packet traffic, IEEE/ACM Transactions on Networking (TON), v.4 n.2, p.209-223, April 1996 [doi>10.1109/90.491008]
• 11: A. Feldmann. Modelling characteristics of tcp connections. Technical report, AT$cT Laboratories, 1996.
• 12: Will E. Leland , Murad S. Taqqu , Walter Willinger , Daniel V. Wilson, On the self-similar nature of Ethernet traffic (extended version), IEEE/ACM Transactions on Networking (TON), v.2 n.1, p.1-15, Feb. 1994 [doi>10.1109/90.282603]
• 13: Bruce A. Mah, An Empirical Model of HTTP Network Traffic, Proceedings of the INFOCOM '97. Sixteenth Annual Joint Conference of the IEEE Computer and Communications Societies. Driving the Information Revolution, p.592, April 09-11, 1997
• 14: R. Mattson, .I. Gecsei, D. Slutz, and I. Traiger. Evaluation techniques and storage hierarchies. IBM Systems Journal, 9:78-117, 1970.
• 15: d.C. Mogul. Network behavior of a busy web server and its clients. Technical Report WRL 95/5, DEC Western Research Laboratory, Palo Alto, CA, 1995.
• 16: University of Minnesota. {(}stone version 1. htt p://web 66. coled, umn .ed u / gsto ne/in fo.html.
• 17: Kihong Park , Gitae Kim , Mark Crovella, On the relationship between file sizes, transport protocols, and self-similar network traffic, Proceedings of the 1996 International Conference on Network Protocols (ICNP '96), p.171, October 29-November 01, 1996
• 18: Vern Paxson, Empirically derived analytic models of wide-area TCP connections, IEEE/ACM Transactions on Networking (TON), v.2 n.4, p.316-336, Aug. 1994 [doi>10.1109/90.330413]
• 19: Shane P. Pederson , Mark E. Johnson, Estimating model discrepancy, Technometrics, v.32 n.3, p.305-314, Aug. 1990 [doi>10.2307/1269107]
• 20: Gene Trent and Mark Sake. Webstone: The first generation in http server benchmarking, February 1995. Silicon (#raphics White Paper.
• 21: Walter Willinger , Murad S. Taqqu , Robert Sherman , Daniel V. Wilson, Self-similarity through high-variability: statistical analysis of Ethernet LAN traffic at the source level, IEEE/ACM Transactions on Networking (TON), v.5 n.1, p.71-86, Feb. 1997 [doi>10.1109/90.554723]
• 22: G. K. Zipf. Human Behavior and the Principle of Least- Effort. Addison-Wesley, Cambridge, MA, 1949.

INDEX TERMS

Primary Classification:
  I. Computing Methodologies
  I.6 SIMULATION AND MODELING

Additional Classification:
  H. Information Systems
  H.3 INFORMATION STORAGE AND RETRIEVAL
      H.3.4 Systems and Software

          Nouns: World Wide Web (WWW)

General Terms:
Measurement, Performance