ABSTRACT
WWW workload generators are used to evaluate web server performance, and thus have a large impact on what performance optimizations are applied to servers. However, current benchmarks ignore a crucial component: how these servers perform in the environment in which they are intended to be used, namely the wide-area Internet.This paper shows how WAN conditions can affect WWW server performance. We examine these effects using an experimental test-bed which emulates WAN characteristics in a live setting, by introducing factors such as delay and packet loss in a controlled and reproducible fashion. We study how these factors interact with the host TCP implementation and what influence they have on web server performance. We demonstrate that when more realistic wide-area conditions are introduced, servers exhibit very different performance properties and scaling behaviors, which are not exposed by existing benchmarks running on LANs. We show that observed throughputs can give misleading information about server performance, and thus find that maximum throughput, or capacity, is a more useful metric. We find that packet losses can reduce server capacity by as much as 50 percent and increase response time as seen by the client. We show that using TCP SACK can reduce client response time, without reducing server capacity.
- 1.Mark Allman. A Web server's view of the transport layer. Computer Communication Review, 30(5), Oct 2000. Google ScholarDigital Library
- 2.Martin F. Arlitt and Carey L. Williamson. Internet Web servers: Workload characterization and performance implications. IEEE/ACM Transactions on Networking, 5(5):631-646, Oct 1997. Google ScholarDigital Library
- 3.Mohit Aron and Peter Druschel. TCP implementation enhancements for improving Webserver performance. Technical Report TR99-335, Rice University Computer Science Dept., July 1999.Google Scholar
- 4.Gaurav Banga and Peter Druschel. Measuring the capacity of a Web server under realistic loads. World Wide Web Journal (Special Issue on World Wide Web Characterization and Performance Evaluation), 2(1):69-83, May 1999. Google ScholarDigital Library
- 5.Gaurav Banga and Jeffrey C. Mogul. Scalable kernel performance for Internet servers under realistic loads. In USENIX Annual Technical Conference, New Orleans, Louisiana, June 1998. Google ScholarDigital Library
- 6.Paul Barford. Modeling, Measurement and Performance of World Wide Web Transactions. PhD thesis, Boston University, Boston, MA 02215, January 2001. Google ScholarDigital Library
- 7.Paul Barford and Mark Crovella. Generating representative Web workloads for network and server performance evaluation. In Proceedings of the ACM Sigmetrics Conference on Measurement and Modeling of Computer Systems, Madison, WI, June 1998. Google ScholarDigital Library
- 8.Paul Barford and Mark Crovella. Critical path analysis of TCP transactions. In ACM SIGCOMM Symposium on Communications Architectures and Protocols, Stockhold, Sweden, August 2000. Google ScholarDigital Library
- 9.Mark Carson. NIST Net. Available at http://www.antd.nist.gov/nistnet.Google Scholar
- 10.The Standard Performance Evaluation Corporation. SpecWeb96/SpecWeb99. http://www.spec.org/osg.Google Scholar
- 11.Mark Crovella and Azer Bestavros. Self-similarity in World Wide Web traffic: Evidence and possible causes. IEEE/ACM Transactions on Networking, 5(6):835-846, Nov 1997. Google ScholarDigital Library
- 12.Jay L. Devore. Probability and Statistics for Engineering and the Sciences (2nd Ed.). Brooks/Cole Publishing Co., 1987.Google Scholar
- 13.Sally Floyd and Kevin Fall. Promoting the use of end-to-end congestion control in the Internet. IEEE/ACM Trans. on Networking, 7(4), August 1999. Google ScholarDigital Library
- 14.Sally Floyd and Tom Henderson. The NewReno modification to TCP's fast recovery algorithm. In Network Information Center RFC 2582 (Experimental), April 1999. Google ScholarDigital Library
- 15.James C. Hu, Irfan Pyarali, and Douglas C. Schmidt. Measuring the impact of event dispatching and concurrency models on Web server performance over high-speed networks. In Proceedings of the 2nd Global Internet Conference (held as part of GLOBECOM '97), Phoenix, AZ, Nov 1997.Google ScholarCross Ref
- 16.Raj Jain. The Art of Computer Systems Performance Analysis. John Wiley & Sons, 1991.Google Scholar
- 17.T. V. Lakshman and U. Madhow. The performance of TCP/IP for networks with high bandwidth-delay products and random loss. IEEE/ACM Trans. on Networking, 5(3), 1997. Google ScholarDigital Library
- 18.Matthew Mathis, Jamshid Mahdavi, Sally Floyd, and Allyn Romanow. TCP selective acknowledgment options. In Network Information Center RFC 2018, October 1996. Google ScholarDigital Library
- 19.Jeffrey C. Mogul. Network behavior of a busy Web server and its clients. Technical Report 95/5, Digital Equipment Corporation Western Research Lab, Palo Alto, CA, October 1995.Google Scholar
- 20.Erich M. Nahum, Tsipora Barzilai, and Dilip Kandlur. Performance issues in WWW servers (extended abstract). In Proceedings of the ACM Sigmetrics Conference on Measurement and Modeling of Computer Systems, Atlanta, GA, May 1999. An extended version has been accepted to appear in IEEE/ACM Transactions on Networking. Google ScholarDigital Library
- 21.Netcraft. The Netcraft WWW server survey. Available at http://www.netcraft.co.uk/Survey.Google Scholar
- 22.Jitendra Padhye, Victor Firoiu, Don Towsley, and Jim Kurose. Modeling TCP throughput: A simple model and its empirical validation. In ACM SIGCOMM Symposium on Communications Architectures and Protocols, September 1998. Google ScholarDigital Library
- 23.Jitendra Padhye and Sally Floyd. Identifying the TCP behavior of Web servers. ICSI Tech Report 01-002, February 2001.Google Scholar
- 24.Vivek Pai, Peter Druschel, and Willy Zwaenepoel. Flash: An efficient and portable Web server. In USENIX Annual Technical Conference, Monterey, CA, June 1999. Google ScholarDigital Library
- 25.Vivek S. Pai, Peter Druschel, and Willy Zwaenepoel. I/O Lite: A copy-free UNIX I/O system. In 3rd USENIX Symposium on Operating Systems Design and Implementation, New Orleans, LA, February 1999.Google Scholar
- 26.Vern Paxson. End-to-end Internet packet dynamics. IEEE/ACM Transactions on Networking, 7(3), June 1999. Google ScholarDigital Library
- 27.Luigi Rizzo. Issues in the implementation of selective acknowledgements for TCP. Draft report. Information available at http://info.iet.unipi.it/luigi/sack.html, Jan 1996.Google Scholar
- 28.Luigi Rizzo. Dummynet: A simple approach to the evaluation of network protocols. Computer Communication Review, 27(2), Feb 1997. Google ScholarDigital Library
- 29.Dan Rubenstein, Jim Kurose, and Don Towsley. Detecting shared congestion of flows via end-to-end measurement. In Proceedings of ACM SIGMETRICS 2000, Santa Clara, CA, June 2000. Google ScholarDigital Library
- 30.Srinivasan Seshan, Hari Balakrishnan, Venkata N. Padmanabhan, Mark Stemm, and Randy Katz. TCP behavior of a busy Internet server: Analysis and improvements. In Proceedings of the Conference on Computer Communications (IEEE Infocom), San Francisco, CA, March 1998.Google Scholar
- 31.W. Richard Stevens. Re: cleaning up TIME WAIT states. In End-to-end mailing list archives, January 1997. Available at http://www.postel.org/mailman/listinfo/end2end-interest.Google Scholar
- 32.Maya Yajnik, Sue B. Moon, Jim Kurose, and Don Towsley. Measurement and modelling of the temporal dependence in packet loss. In Proceedings of 1999 IEEE INFOCOM, New York, NY, March 1999.Google ScholarCross Ref
- The effects of wide-area conditions on WWW server performance
Recommendations
Enhancing TCP performance in wide-area cellular wireless networks: transport level approaches
Wireless communications systems and networksInternet technology-based architectures and protocols for supporting multimedia traffic over wireless networks are evolving. Since TCP (Transmission Control Protocol)/IP (Internet Protocol) is the standard network protocol stack on the Internet, its use ...
The effects of wide-area conditions on WWW server performance
WWW workload generators are used to evaluate web server performance, and thus have a large impact on what performance optimizations are applied to servers. However, current benchmarks ignore a crucial component: how these servers perform in the ...
On the performance of wide-area thin-client computing
While many application service providers have proposed using thin-client computing to deliver computational services over the Internet, little work has been done to evaluate the effectiveness of thin-client computing in a wide-area network. To assess ...
Comments