Abstract
An extensive set of measurements were made in an OSI testbed to study the behavior of congestion control and avoidance. Testbed systems used the Connectionless Network Protocol (CLNP) and Transport Protocol Class 4 (TP4), which had been modified to perform the CE-bit [10] congestion avoidance and the "CUTE" [6] congestion recovery algorithms.We found that two-way traffic has dynamics which can significantly decrease fairness among competing connections using congestion avoidance. We present experiments that demonstrate this problem and our analysis of how two-way traffic results in reduced fairness. This analysis led us to develop an effective modification to the congestion avoidance algorithms to maintain fairness with two-way traffic.Our analysis of experimental results also points to undesirable interactions between two-way traffic dynamics and a sending strategy that times data transmissions, by the receipt of acknowledgements. These interactions reinforce burstiness of transmissions. therefore increasing buffering requirements and delay in routers. They may also decrease throughput.
- {1} T. Boland, "Stable Implementation Agreements for OSI Protocols", Vol. 3, Ed. 1, Dec. 1989, NIST/SP/500/177, U. S. Dept. of Commerce, National Institute of Science and Technology.Google Scholar
- {2} J. Bolot and A. U. Shankar, "Dynamical Behavior of Rate-Based Flow Control Mechanisms", Computer Communications Review, Vol. 20, No. 2, April 1990, pp. 35-49. Google ScholarDigital Library
- {3} Bux, W. and Grillo, D. "Flow Control in Local Area Networks of Interconnected Token Rings", IEEE Transactions on Communication, COM-33, 10, Oct. 1985, pp. 1058-1066.Google Scholar
- {4} V. Jacobson, "Congestion Avoidance and Control", Proc. ACM SIGCOMM '88, Computer Communications Review, Vol. 18, No. 4, August 1988, pp. 314-329. Google ScholarDigital Library
- {5} V. Jacobson, "Berkeley TCP Evolution from 4.3-tahoe to 4.3-reno", IETF Proceedings, Vancouver, B. C., August 1990.Google Scholar
- {6} R. Jain, "Timeout-based Congestion Control Scheme for Window-controlled Networks", IEEE Journal on Selected Areas in Communications, SAC-4, 7, Oct. 1986, pp. 1162-1167.Google ScholarDigital Library
- {7} A. Mankin, "Random Drop Congestion Control", Proc. ACM SIGCOMM '90, Computer Communications Review, Vol. 20, No. 4, Sept. 1990, pp. 1-7. Google ScholarDigital Library
- {8} A. Mankin, et al. "Evaluation of Internet Performance - FY89", Technical Report MTR-89W00216, The MITRE Corp, McLean, VA.Google Scholar
- {9} D. Mitra and J. B. Seery, "Dynamic Adaptive Windows for High Speed Data Networks: Theory and Simulations", Proc. ACM SIGCOMM '90, Computer Communications Review, Vol. 20, No. 4, Sept. 1990, pp. 30-40. Google ScholarDigital Library
- {10} K. K. Ramakrishnan and R. Jain, "A Binary Feedback Scheme for Congestion Avoidance in Computer Networks with a Connectionless Network Layer", Proc. ACM SIGCOMM '88, Computer Communications Review, Vol. 18, No. 4, Aug. 1988, pp. 303-313. Google ScholarDigital Library
- {11} S. Shenker, "A Theoretical Analysis of Feedback Flow Control", Proc. ACM SIGCOMM '90, Computer Communications Review, Vol. 20, No. 4, Sept. 1990, pp. 156-165. Google ScholarDigital Library
- {12} S. Shenker, L. Zhang, and D.D. Clark, "Dynamics of a Congestion Control Algorithm", Computer Communications Review, Vol. 20, No. 5, Oct. 1990, pp. 30-39. Google ScholarDigital Library
- {13} Z. Wang and J. Crowcroft, "A New Congestion Control Scheme: Slow Start and Search (Tri-S)", Computer Communication Review, Volume 21, No. 1, pp. 32- 43, Jan. 1991. Google ScholarDigital Library
- {14} L. Zhang, "Virtual Clock: A New Traffic Control Algorithm for Packet Switching Networks", Proc. ACM SIGCOMM '90, Computer Communications Review, Vol. 20, No. 4, Sept. 1990, pp. 19-29. Google ScholarDigital Library
Index Terms
- Dynamics of congestion control and avoidance of two-way traffic in an OSI testbed
Recommendations
Delay-based congestion avoidance for TCP
The set of TCP congestion control algorithms associated with TCP/Reno (e.g., slow-start and congestion avoidance) have been crucial to ensuring the stability of the Internet. Algorithms such as TCP/NewReno (which has been deployed) and TCP/Vegas (which ...
Congestion avoidance and control
SIGCOMM '88: Symposium proceedings on Communications architectures and protocolsIn October of '86, the Internet had the first of what became a series of 'congestion collapses'. During this period, the data throughput from LBL to UC Berkeley (sites separated by 400 yards and three IMP hops) dropped from 32 Kbps to 40 bps. Mike ...
Comments