A Window-Based Retransmission Timeout for TCP

We present a new timeout algorithm for TCP, based on the observation that TCP-RTO should not be solely based on RTT estimations. We argue that the design principles of the current timeout algorithm may lead to flow synchronization, unnecessary retransmission effort and unfair resource allocation. WB-RTO exhibits two major properties: (i) it cancels retrans- mission synchronization which dominates when resource demand exceeds by far resource supply and (ii) reschedules flows on the basis of their contribution to congestion. I. INTRODUCTION The retransmission timeout policy of standard TCP (15) is governed by the rules of RFC 2988 (14). The algorithm is based solely on RTT measurements, trying to capture dynamic network conditions by measuring the variation of the RTT samples. In particular, the Retransmission Timeout is calculated upon each ACK arrival after smoothing out the measured samples, and weighting the recent history. More precisely, upon each ACK arrival, the sender:  calculates the RTT Variation: RTTV AR =3 /4 · RTTV AR+

[1]  Van Jacobson,et al.  TCP Extensions for High Performance , 1992, RFC.

[2]  Andrei Gurtov,et al.  Evaluating the Eifel Algorithm for TCP in a GPRS Network , 2001 .

[3]  Randy H. Katz,et al.  The Eifel algorithm: making TCP robust against spurious retransmissions , 2000, CCRV.

[4]  Liang Guo,et al.  The war between mice and elephants , 2001, Proceedings Ninth International Conference on Network Protocols. ICNP 2001.

[5]  W. Richard Stevens,et al.  TCP Slow Start, Congestion Avoidance, Fast Retransmit, and Fast Recovery Algorithms , 1997, RFC.

[6]  QUTdN QeO,et al.  Random early detection gateways for congestion avoidance , 1993, TNET.

[7]  Vern Paxson,et al.  TCP Congestion Control , 1999, RFC.

[8]  Kimmo E. E. Raatikainen,et al.  F-RTO: an enhanced recovery algorithm for TCP retransmission timeouts , 2003, CCRV.

[9]  Reiner Ludwig,et al.  Responding to spurious timeouts in TCP , 2003, IEEE INFOCOM 2003. Twenty-second Annual Joint Conference of the IEEE Computer and Communications Societies (IEEE Cat. No.03CH37428).

[10]  Mark Allman,et al.  On making TCP more robust to packet reordering , 2002, CCRV.

[11]  Ioannis Psaras,et al.  Why TCP timers (still) don't work well , 2007, Comput. Networks.

[12]  Lixia Zhang,et al.  Why TCP timers don't work well , 1986, SIGCOMM '86.

[13]  Vern Paxson,et al.  Computing TCP's Retransmission Timer , 2000, RFC.

[14]  Vern Paxson,et al.  On estimating end-to-end network path properties , 2001, SIGCOMM LA '01.

[15]  Reiner Ludwig,et al.  The peak-hopper: a new end-to-end retransmission timer for reliable unicast transport , 2004, IEEE INFOCOM 2004.

[16]  Ioannis Psaras,et al.  CA-RTO: a contention-adaptive retransmission timeout , 2005, Proceedings. 14th International Conference on Computer Communications and Networks, 2005. ICCCN 2005..

[17]  Raj Jain,et al.  Analysis of the Increase and Decrease Algorithms for Congestion Avoidance in Computer Networks , 1989, Comput. Networks.