Performance issues of enterprise level web proxies

ABSTRACT

Enterprise level web proxies relay world-wide web traffic between private networks and the Internet. They improve security, save network bandwidth, and reduce network latency. While the performance of web proxies has been analyzed based on synthetic workloads, little is known about their performance on real workloads. In this paper we present a study of two web proxies (CERN and Squid) executing real workloads on Digital's Palo Alto Gateway. We demonstrate that the simple CERN proxy architecture outperforms all but the latest version of Squid and continues to outperform cacheless configurations. For the measured load levels the Squid proxy used at least as many CPU, memory, and disk resources as CERN, in some configurations significantly more resources. At higher load levels the resource utilization requirements will cross and Squid will be the one using fewer resources. Lastly we found that cache hit rates of around 30% had very little effect on the requests service time.

References

1. 1.J ussara Almeida, Virgilio Almeida, and David Yates. Measuring the Behavior of a World-Wide Web Server. Technical Report CS 96-025, Boston University, October 29 1996.]]
2. 2.Virgilio Almeida, Azer Bestavros, Mark Crovella, , and Adria#a de Oliveira. Characterizing Reference Locality in the WWW. In IEEE PDIS'96: The International Conference in Parallel and Distributed information Systems, Miami Beach, Florida, December 1996. IEEE.]]
3. 3.Martin F. Arlitt and Carey L. Williamson. Web Server Workload Characterization: The Search for Invariants. In ACM Sigmetrics '96, pages 126-137, Philadelphia, PA, May 23-26 1996. ACM Sigmetrics, ACM.]]
4. 4.Lance Berc, Sanjay Ghemawat, Monika Henzinger, Shun-Tak Leung, Mitch Lichtenberg, Dick Sites, Mark Vandevoorde, Carl Waldspurger, and Bill Weihl. DIGITAL Continuous Profiling Infrastructure. Available on the World Wide Web at http ://www.research.digital.com/SR C/dcvi/p#pe#/o#di96- wip.html, October 1996.]]
5. 5.C. Mic Bowman, Peter B. Danzig, Darren R. Hardy, Udi Man ber, Michael F. Schwartz, and Duane P. Wessels. Harvest: A scalable, customizable discovery and access system. Technical Report CU-CS-732-94, Department of Computer Science, University of Colorado, Boulder , CO, August 1994 (revised March 1995) 1994.]]
6. 6.Anawat Chankhunthod, Peter B. Danzig, Chuck Neerdaels, Michael F. Schwartz, and Kurt J. Worrell. A Hierarchical Internet Object Cache. In 1996 USENIX Technical Conference, San Diego, CA, January 1996. USENIX.]]
7. 7.Anawat Chankhunthod, Peter B. Danzig, Chuck Neerdaeis, Duane Wessels, Mike F. Schwartz, and Erhyuan Tsai. The Harvest Cache and Httpd- Accelerator. Available on the World Wide Web at http://ezcalibur.usc.edu/, July 1995.]]
8. 8.Dan Connolly and Tim Berners-Lee. Names and Addresses, URIs, URLs, URNs, URCs. Available on the World Wide Web at http://www.w3.org/pub/WWW/Addressing/, December 1990.]]
9. 9.M. Crovella and A. Bestavros. Self-similarity in World- Wide Web Traffic: Evidence and Possible Causes. In Proc. of the 1996 SIGMETRICS Conference on Measurement and Modeling of Computer Systems, 1996, May 1996. ACM.]]
10. 10.Peter Druschel and Gaurav Bangs. Lazy Reiceiver Processing (LRP): A Network Subsystem Architecture for Server Systems. In Second Symposium on Operating System Design and Implementation (OSDI 96}, Seattle, WA, October 1996.]]
11. 11.Thomas Kroeger and Jeffrey Mogul. Digital's Web Proxy Traces. Available on the World Wide Web at flp ://tip. digital, com/pub/D E C/traces/proxy/webtraces.html, October 1996.]]
12. 12.hri Luotonen, Henrik Frystyk Nielsen, and Tim Berners-Lee. CERN httpd 3.0A. Available on the World Wide Web at http://www.wS.org/pub/WWW/Daemon/, July 15 1996.]]
13. 13.P. Mockapetris. Domain Names- Concepts and Facilities. Available on the World Wide Web at tip ://flp. internic, net/rfc/rf c 1034.txt, November 1987.]]
14. 14.P. Mockapetris. Domain Names- Implementation and Specification. Available on the World Wide Web at flp://flp.internic.net/rfc/rfc1035.txt, November 1987.]]
15. 15.Jeffrey C. Mogul. Network behavior of a busy web server and its clients. Technical Report 95/5, DEC Western Research Laboratory, Palo Alto, CA, October 1995.]]
16. 16.Jeffrey C. Mogul. Operating Systems Support for Busy Internet Servers. In Fifth Workshop on Hot Topics in Operating Systems (HotOS- V), page addendum, Orcas Island, Washington, May 1995.]]
17. 17.Jeffrey C. Mogul and K. K. Ramakrishnan. Ehminating receive livelock in an interrupt-driven kernel. In 1996 Usenix Technical Gon}erence, pages 99-111, San Diego, CA, January 1996.]]
18. 18.Standard Performance Evaluation Corp. (SPEC). SPECweb96 Benchmark. Available on the World Wide Web at http://www.specbench.org/osg/web96/, July 1996.]]
19. 19.Gene Trent and Mark Sake. WebSTONE: The First Generation in HTTP Server Benchmarking. Available on the World Wide Web at http://www.sgi.com/Products/Web FOR CE/ WebStone/, February 1995.]]
20. 20.Neal R. Wagner. Fingerprinting. In Symposium on Security and Privacy, pages 18-22, Oakland, CA, 1983. IEEE.]]
21. 21.Duane Wesseis. Personal communication, during which became clear that Squid 1.0.betal7 has a memory management bug, August 1996.]]
22. 22.Duane Wessels. Squid Internet Object Cache. Available on the World Wide Web at http://squid.nlanr.net/Squid/, May 1996.]]
23. 23.Duane Wessels. SQUID Frequently Asked Questions. Available on the World Wide Web at http://squid.nlanr.net/Squid/FA Q.html, January 1997.]]

Index Terms

• Computer systems organization

  • Dependable and fault-tolerant systems and networks

• General and reference

  • Cross-computing tools and techniques

    • Performance

• Networks

  • Network architectures

  • Network performance evaluation