Web100: extended TCP instrumentation for research, education and diagnosis

ABSTRACT

TCP has become the dominant protocol for all network data transport because it presents a simple uniform data delivery service which is sufficient for most applications over all types of lower network layers. By its very nature, TCP's adaption and retransmission strategies hide all of the details of the lower layers from the application. For example the only symptom of spurious packet loss (or nearly any other network problem) is longer elapsed time and lower performance.This information hiding is fundamentally important to the growth of the Internet because it decouples the evolution of applications from the evolution of link layers. However it also hides valuable information from researchers, educators, network administrators, and other people who would benefit from insight into the inner workings of TCP and the lower layers.In this paper, we present an architecture and infrastructure that provides for per-connection TCP instrumentation to expose otherwise hidden protocol events. We show examples how the infrastructure can be used in support of research, education and advanced network diagnostic tools.Our work was motivated by the observation that since about 1985 network data rates for typical novice network users have fallen by about three orders of magnitude behind expert users (who have kept up with Moore's Law). We use the term "Wizard Gap" to describe this phenomenon. The Web100 and Net100 projects were formed as one step in closing the Wizard Gap.

REFERENCES

• 1: M. Allman, V. Paxson, and W. Stevens. Tcp congestion control, RFC2581, April 1999.
• 2: ANINEAR. Advanced networking infrastructure needs in the atmospheric and related sciences (aninars) workshop report. http://www.scd.ucar.edu/nets/projects/completed/1999.complete.projects/nlanr/ final.report.htm.
• 3: Lawrence S. Brakmo , Sean W. O'Malley , Larry L. Peterson, TCP Vegas: new techniques for congestion detection and avoidance, Proceedings of the conference on Communications architectures, protocols and applications, p.24-35, August 31-September 02, 1994, London, United Kingdom
• 4: CAIDA. Internet tools taxonomy, 2003. http://www.caida.org/tools/taxonomy/.
• 5: R. Carlson. Developing the Web100 based network diagnostic tool (NDT). PAM, April 2003.
• 6: D. D. Clark. Window and acknowledgement strategy in TCP, RFC813, July 1982.
• 7: T. Dunigan. Floyd's TCP slow-start and AIMD mods, 2003. http://www.csm.ornl.gov/~dunigan/netperf/floyd.html.
• 8: T. Dunigan. Kelly's scalable TCP AIMD mods, 2003. http://www.csm.ornl.gov/~dunigan/netperf/kelly.html.
• 9: T. Dunigan. ORNL TCP Web100 bandwidth tester, 2003. http://firebird.ccs.ornl.gov:7123/.
• 10: Tom Dunigan , Matt Mathis , Brian Tierney, A TCP tuning daemon, Proceedings of the 2002 ACM/IEEE conference on Supercomputing, p.1-16, November 16, 2002, Baltimore, Maryland
• 11: Wu-chun Feng , Mike Fisk , Mark K. Gardner , Eric Weigle, Dynamic Right-Sizing: An Automated, Lightweight, and Scalable Technique for Enhancing Grid Performance, Proceedings of the 7th IFIP/IEEE International Workshop on Protocols for High Speed Networks, p.69-83, April 22-24, 2002
• 12: S. Floyd. HighSpeed TCP for Large Congestion Windows. Work-in- Progress: IETF Internet-Draft, August 2003. http://www.ietf.org/internetdrafts/draft-ietf-tsvwg-highspeed-01.txt.
• 13: S. Floyd. Limited Slow-Start for TCP with Large Congestion Windows. Work in progress: IETF Internet-Draft, July 2003. http://www.ietf.org/internet-drafts/draft-ietf-tsvwg-slowstart-00.txt.
• 14: Sally Floyd , Vern Paxson, Difficulties in simulating the internet, IEEE/ACM Transactions on Networking (TON), v.9 n.4, p.392-403, August 2001 [doi>10.1109/90.944338]
• 15: M. Handley, J. Padhye, and S. Floyd. TCP congestion window validation, RFC2861, June 2000.
• 16: J. Heffner. High bandwidth TCP queuing, July 2002. http://www.psc.edu/~jheffner/papers/senior_thesis.ps.
• 17: V. Jacobson. Modified TCP congestion avoidance algorithm. Message to end2end-interest list, April 1990. ftp://ftp.ee.lbl.gov/email/vanj.90apr30.txt.
• 18: V. Jacobson, R. Braden, and D. Borman. TCP extensions for high performance, RFC1323, May 1992.
• 19: C. Jin et al. FAST kernel: Background theory and experimental results. In First International Workshop on Protocols for Fast Long-Distance Networks, February 2003.
• 20: T. Kelly. Scalable TCP: Improving performance in highspeed wide area networks. In First International Workshop on Protocols for Fast Long-Distance Networks, February 2003.
• 21: LBNL. Network tools analysis framework (ntaf), 2003. http://www-didc.lbl.gov/NTAF/.
• 22: Jason Lee , Dan Gunter , Martin Stoufer , Brian Tierney, Monitoring data archives for grid environments, Proceedings of the 2002 ACM/IEEE conference on Supercomputing, p.1-10, November 16, 2002, Baltimore, Maryland
• 23: M. Mathis. Pushing up performance for everyone, December 1999. Presentation to Joint Techs workshop (first use of wizard gap).
• 24: M. Mathis, J. Heffner, R. Reddy, and J. Saperia. TCP Extended Statistics MIB. Work in progress: IETF Internet-Draft, November 2002. Status page: http://www.web100.org/mib.
• 25: M. Mathis, J. Mahdavi, S. Floyd, and A. Romanow. TCP selective acknowledgement options, RFC2018, October 1996.
• 26: M. Mathis and R. Reddy. Enabling High Performance Data Transfers, 2002. http://www.psc.edu/networking/perf_tune.html.
• 27: M. Mathis and R. Reddy. Pathprobe: Network Path Diagnostic Tools, 2002. http://www.psc.edu/~web100/pathprobe/.
• 28: Matthew Mathis , Jeffrey Semke , Jamshid Mahdavi, The macroscopic behavior of the TCP congestion avoidance algorithm, ACM SIGCOMM Computer Communication Review, v.27 n.3, p.67-82, July 1997 [doi>10.1145/263932.264023]
• 29: J. Mogul and S. Deering. Path MTU discovery, RFC1191, November 1990.
• 30: J. Nagle. Congestion control in IP/TCP internetworks, RFC896, January 1984.
• 31: Net100. Home page, 2003. http://www.net100.org/.
• 32: NLANR. Iperf---the TCP/UDP bandwidth measurement tool, 2002. http://dast.nlanr.net/Projects/Iperf/.
• 33: S. Ostermann. TCPtrace, 2003. http://www.tcptrace.org/.
• 34: V. Paxson and M. Allman. Computing TCP's retransmission timer, RFC2988, November 2000.
• 35: Vern Paxson , Sally Floyd, Why we don't know how to simulate the Internet, Proceedings of the 29th conference on Winter simulation, p.1037-1044, December 07-10, 1997, Atlanta, Georgia, United States [doi>10.1145/268437.268737]
• 36: K. Ramakrishnan, S. Floyd, and D. Black. A proposal to add explicit congestion notification (ECN) to IP, RFC3168, September 2001.
• 37: R. Reddy. SYN option check server, 2003. http://syntest.psc.edu:7961/.
• 38: Jeffrey Semke , Jamshid Mahdavi , Matthew Mathis, Automatic TCP buffer tuning, Proceedings of the ACM SIGCOMM '98 conference on Applications, technologies, architectures, and protocols for computer communication, p.315-323, August 31-September 04, 1998, Vancouver, British Columbia, Canada
• 39: B. Tierney. Using NetLogger and Web100 for TCP analysis. Protocols for High Speed Networks. http://www-didc.lbl.gov/papers/PFDL.tierney.pdf.
• 40: A. Tirumala, L. Cottrell, and T. Dunigan. Measuring end-toend bandwidth with Iperf using Web100. PAM, April 2003.
• 41: G. Turner. tcpestats: A Net-SNMP AgentX agent implementing the Web100 Project's TCP Extended Statistics MIB, 2002. http://www.aarnet.edu.au/network/software/web100/.
• 42: Curtis Villamizar , Cheng Song, High performance TCP in ANSNET, ACM SIGCOMM Computer Communication Review, v.24 n.5, p.45-60, Oct. 1994 [doi>10.1145/205511.205520]
• 43: Web100. Kernel Instrument Set, 2002. http://www.web100.org/download/kernel/alpha2.0/tcp-kis.txt.
• 44: G. Wood. 12--NEWS: Internet2 Land Speed Winners Set New Transcontinental Internet Performance Records, 2002. http://mail.internet2.edu:8080/guest/archives/i2-news/log200003/msg00011.html.

INDEX TERMS

Keywords:
Net100, TCP Performance, Web100, instrumentation