ABSTRACT
Repeated observation of TCP retransmission timer problems stimulated investigation into the roles and limitations of timers. Timers are indispensable tools in building up reliable distributed systems. However, as the experience with the TCP retransmission timer has shown, timers have intrinsic limitations in offering optimal performance. Any timeout based action is a guess based on incomplete information, and as such is bound to be non-optimal. We conclude that, if we aim at high performance, we should use external events as a first line of defense against failures, and depend on timers only in cases where external notification has failed.
- 1.Davld Clark. ~Vinclow and Acknowledgment Strategy in TC, P. ARPA RFC-813. 1,082]] Google ScholarDigital Library
- 2.David Clarlc, Mark Lambert, & bixia Zhang. NETBLT: A Bull< Data Transfer Protocol. ARPA RFC-069. December, 198.5]]Google Scholar
- 3.Geoffrey Cooper. A New Timing A.lgoritllnl for Transmission and Retransmission in TFTP. A working paper dra.f'l, written a,t Computer System Research group, MIT-LCS. t0sa]]Google Scholar
- 4.St,epllen \V. Edge. An Adaptive Timeout Algorithm for Retransmission Across a Packet Switching Network. ACA/Comp,.ter Com,~u~icatioa Review 14(2):248-255, June, 198.'t.]] Google ScholarDigital Library
- 5.1}a.} Jain. Di.verge,ace of 2:i.m..eoul, Algorithms for Packet Ret./'a./~ 8.~ ,.i.s.s i o n,. Technical Report 329, Digital Equipment Corp., 1985.]]Google Scholar
- 6.David L. Mills. lnl,ernet Delay Experiments. AR PA RFC-880. Decem l>er, 1083]]Google Scholar
- 7.Robert .J.T. ~{oB'ris. Fixing Titneo~lt lnt.ervals for Lost Packet Detect, ion in Computer Communication Networks. In P~'oc. o.f N, tio~,,l C'om.pltter Conference. AFIPS, 197.0.]]Google Scholar
- 8..I. Postel. DoD Standard Transmission Control Protocol. AI~PA RFC-793. S~'pt, c11}ber. 1981]]Google Scholar
- 9.Postel. DoD Sta. nda~'d Internet P!'otoeol. AFCPA RI:'C-70I. September, 1981]]Google Scholar
- 10.Lixin, Zltang. Net. work Sinlttlat,iot~ Report. \Vorking paper in progress. This ,'eport summarizes test, results on IP source quenell handling and TCP timer problems. The simulator wa.s built by the autllor a.t MIT-bCS to study network eonp:;estion control problems. Its topology imitates the conditions in the current ARPA lnternet, i.e. the delay and l)m~dwidl,h characteristics of colnmunication cl~annels differ by orders of magnil, ude. T}~e data traffic generator models two tyt>es of' applications: file transfer and renlo(.e login,]]Google Scholar
Index Terms
- Why TCP timers don't work well
Recommendations
Why TCP timers don't work well
Repeated observation of TCP retransmission timer problems stimulated investigation into the roles and limitations of timers. Timers are indispensable tools in building up reliable distributed systems. However, as the experience with the TCP ...
Why TCP timers (still) don't work well
We argue that the design principles of the TCP timeout algorithm are based solely on RTT estimations and may lead to flow synchronization, unnecessary retransmission effort and unfair resource allocation. We present a new Window-Based Retransmission ...
TCP CERL: congestion control enhancement over wireless networks
In this paper, we propose and verify a modified version of TCP Reno that we call TCP Congestion Control Enhancement for Random Loss (CERL). We compare the performance of TCP CERL, using simulations conducted in ns-2, to the following other TCP variants: ...
Comments