Evaluation of TCP congestion control algorithms.

Abstract

Sandia, Los Alamos, and Lawrence Livermore National Laboratories currently deploy high speed, Wide Area Network links to permit remote access to their Supercomputer systems. The current TCP congestion algorithm does not take full advantage of high delay, large bandwidth environments. This report involves evaluating alternative TCP congestion algorithms and comparing them with the currently used congestion algorithm. The goal was to find if an alternative algorithm could provide higher throughput with minimal impact on existing network traffic. The alternative congestion algorithms used were Scalable TCP and HighSpeed TCP. Network lab experiments were run to record the performance of each algorithm under different network configurations. The network configurations used were back-to-back with no delay, back-to-back with a 30ms delay, and two-to-one with a 30ms delay. The performance of each algorithm was then compared to the existing TCP congestion algorithm to determine if an acceptable alternative had been found. Comparisons were made based on throughput, stability, and fairness.

Network protocol changes can improve DisCom WAN performance : evaluating TCP modifications and SCTP in the ASC tri-lab environment.

“…The figures below, extracted from previous works [1,2], show that the increased (best case of 12%-to 5%) from standard TCP currently does not warrant the modification to the TCP stack. …”

“…However, none of these proposals has been adopted into the main stream TCP implementations. A laboratory version of the DisCom WAN was created so that Sandia could investigate these modifications, evaluate them in great detail and report their performance [1,2]. In addition to modifications to TCP a new protocol, the Stream Control Transmission Protocol (SCTP), had been developed and become part of the Linux operating system environment.…”

Evaluation of replacement protocols and modifications to TCP to enhance ASC Wide Area Network performance.

“…For definitions, equations, and extensive research on the Standard TCP control algorithm, see the SAND2003-4404 document [3]. The SAND2003-4404 document demonstrates the evaluation of the Standard TCP algorithms used such as: slow start, congestion avoidance, fast retransmit, and fast recovery.…”

“…A goal of HighSpeed TCP is to behave similarly to Standard TCP when using small congestion windows. For definitions, equations, and extensive research on the High-Speed TCP control algorithm, see the SAND2003-4404 document [3].…”

“…Scalable TCP is a simple change to the traditional TCP congestion control algorithm (RFC2581), which dramatically improves TCP performance in high-speed wide area networks [13]. For definitions, equations, and extensive research on the Scalable TCP control algorithm, see the SAND2003-4404 document [3].…”