TCP congestion control
Download as PPT, PDF12 likes11,174 views
This document provides an overview of TCP congestion control algorithms. It describes the basic additive increase/multiplicative decrease approach and key mechanisms like slow start, fast retransmit, and fast recovery. It also discusses algorithms for setting the retransmission timeout value and adaptations made in protocols like New Reno and Cubic.
TCP congestion control
- 1. TTCCPP CCoonnggeessttiioonn CCoonnttrrooll Lecture material taken from “Computer Networks A Systems Approach”, Fourth Edition,Peterson and Davie, Morgan Kaufmann, 2007. Computer Networks: TTCCPP CCoonnggeessttiioonn CCoonnttrrooll 11
- 2. TTCCPP CCoonnggeessttiioonn CCoonnttrrooll • Essential strategy :: The TCP host sends packets into the network without a reservation and then the host reacts to observable events. • Originally TCP assumed FIFO queuing. • Basic idea :: each source determines how much capacity is available to a given flow in the network. • ACKs are used to ‘pace’ the transmission of packets such that TCP is “self-clocking”. Computer Networks: TTCCPP CCoonnggeessttiioonn CCoonnttrrooll 22
- 3. AAIIMMDD ((AAddddiittiivvee IInnccrreeaassee // MMuullttiipplliiccaattiivvee DDeeccrreeaassee)) • CongestionWindow (cwnd) is a variable held by the TCP source for each connection. MaxWindow :: min (CongestionWindow , AdvertisedWindow) EffectiveWindow = MaxWindow – (LastByteSent -LastByteAcked) • cwnd is set based on the perceived level of congestion. The Host receives implicit (packet drop) or explicit (packet mark) indications of internal congestion. Computer Networks: TTCCPP CCoonnggeessttiioonn CCoonnttrrooll 33
- 4. AAddddiittiivvee IInnccrreeaassee ((AAII)) • Additive Increase is a reaction to perceived available capacity (referred to as congestion avoidance stage). • Frequently in the literature, additive increase is defined by parameter α (where the default is α = 1). • Linear Increase :: For each “cwnd’s worth” of packets sent, increase cwnd by 1 packet. • In practice, cwnd is incremented fractionally for each arriving ACK. increment = MSS x (MSS /cwnd) cwnd = cwnd + increment Computer Networks: TTCCPP CCoonnggeessttiioonn CCoonnttrrooll 44
- 5. Source Destination Add one packet each RTT FFiigguurree 66..88 AAddddiittiivvee IInnccrreeaassee Computer Networks: TTCCPP CCoonnggeessttiioonn CCoonnttrrooll 55
- 6. MMuullttiipplliiccaattiivvee DDeeccrreeaassee ((MMDD)) * Key assumption :: a dropped packet and resultant timeout are due to congestion at a router. • Frequently in the literature, multiplicative decrease is defined by parameter β (where the default is β = 0.5) Multiplicate Decrease:: TCP reacts to a timeout by halving cwnd. • Although defined in bytes, the literature often discusses cwnd in terms of packets (or more formally in MSS == Maximum Segment Size). • cwnd is not allowed below the size of a single packet. Computer Networks: TTCCPP CCoonnggeessttiioonn CCoonnttrrooll 66
- 7. AAIIMMDD ((AAddddiittiivvee IInnccrreeaassee // MMuullttiipplliiccaattiivvee DDeeccrreeaassee)) • It has been shown that AIMD is a necessary condition for TCP congestion control to be stable. • Because the simple CC mechanism involves timeouts that cause retransmissions, it is important that hosts have an accurate timeout mechanism. • Timeouts set as a function of average RTT and standard deviation of RTT. • However, TCP hosts only sample round-trip time once per RTT using coarse-grained clock. Computer Networks: TTCCPP CCoonnggeessttiioonn CCoonnttrrooll 77
- 8. FFiigguurree 66..99 TTyyppiiccaall TTCCPP SSaawwttooootthh PPaatttteerrnn Computer Networks: TTCCPP CCoonnggeessttiioonn CCoonnttrrooll 88 60 50 40 20 1.0 2.0 3.0 4.0 5.0 6.0 7.0 8.0 9.0 Time (seconds) 70 30 10 10.0
- 9. SSllooww SSttaarrtt • Linear additive increase takes too long to ramp up a new TCP connection from cold start. • Beginning with TCP Tahoe, the slow start mechanism was added to provide an initial exponential increase in the size of cwnd. Remember mechanism by: slow start prevents a slow start. Moreover, slow start is slower than sending a full advertised window’s worth of packets all at once. Computer Networks: TTCCPP CCoonnggeessttiioonn CCoonnttrrooll 99
- 10. SSllooww SSttaarrtt • The source starts with cwnd = 1. • Every time an ACK arrives, cwnd is incremented. cwnd is effectively doubled per RTT “epoch”. • Two slow start situations: At the very beginning of a connection {cold start}. When the connection goes dead waiting for a timeout to occur (i.e, the advertized window goes to zero!) Computer Networks: TTCCPP CCoonnggeessttiioonn CCoonnttrrooll 1100
- 11. Source Destination Slow Start Add one packet per ACK FFiigguurree 66..1100 SSllooww SSttaarrtt Computer Networks: TTCCPP CCoonnggeessttiioonn CCoonnttrrooll 1111
- 12. SSllooww SSttaarrtt • However, in the second case the source has more information. The current value of cwnd can be saved as a congestion threshold. • This is also known as the “slow start threshold” ssthresh. Computer Networks: TTCCPP CCoonnggeessttiioonn CCoonnttrrooll 1122
- 13. ssthresh Computer Networks: TTCCPP CCoonnggeessttiioonn CCoonnttrrooll 1133
- 14. FFiigguurree 66..1111 BBeehhaavviioorr ooff TTCCPP CCoonnggeessttiioonn CCoonnttrrooll Computer Networks: TTCCPP CCoonnggeessttiioonn CCoonnttrrooll 1144 60 50 40 20 1.0 2.0 3.0 4.0 5.0 6.0 7.0 8.0 9.0 Time (seconds) 70 30 10
- 15. FFaasstt RReettrraannssmmiitt • Coarse timeouts remained a problem, and Fast retransmit was added with TCP Tahoe. • Since the receiver responds every time a packet arrives, this implies the sender will see duplicate ACKs. Basic Idea:: use duplicate ACKs to signal lost packet. Fast Retransmit Upon receipt of three duplicate ACKs, the TCP Sender retransmits the lost packet. Computer Networks: TTCCPP CCoonnggeessttiioonn CCoonnttrrooll 1155
- 16. FFaasstt RReettrraannssmmiitt • Generally, fast retransmit eliminates about half the coarse-grain timeouts. • This yields roughly a 20% improvement in throughput. • Note – fast retransmit does not eliminate all the timeouts due to small window sizes at the source. Computer Networks: TTCCPP CCoonnggeessttiioonn CCoonnttrrooll 1166
- 17. Sender Receiver ACK 2 FFiigguurree 66..1122 FFaasstt RReettrraannssmmiitt Computer Networks: TTCCPP CCoonnggeessttiioonn CCoonnttrrooll 1177 Packet 1 Packet 2 Packet 3 Packet 4 Packet 5 Packet 6 Retransmit packet 3 ACK 1 ACK 2 ACK 2 ACK 2 ACK 6 Fast Retransmit Based on three duplicate ACKs
- 18. FFiigguurree 66..1133 TTCCPP FFaasstt RReettrraannssmmiitt TTrraaccee Computer Networks: TTCCPP CCoonnggeessttiioonn CCoonnttrrooll 1188 60 50 40 20 1.0 2.0 3.0 4.0 5.0 6.0 7.0 Time (seconds) 70 30 10
- 19. FFaasstt RReeccoovveerryy • Fast recovery was added with TCP Reno. • Basic idea:: When fast retransmit detects three duplicate ACKs, start the recovery process from congestion avoidance region and use ACKs in the pipe to pace the sending of packets. Fast Recovery After Fast Retransmit, half cwnd and commence recovery from this point using linear additive increase ‘primed’ by left over ACKs in pipe. Computer Networks: TTCCPP CCoonnggeessttiioonn CCoonnttrrooll 1199
- 20. MMooddiiffiieedd SSllooww SSttaarrtt • With fast recovery, slow start only occurs: –At cold start –After a coarse-grain timeout • This is the difference between TCP Tahoe and TCP Reno!! Computer Networks: TTCCPP CCoonnggeessttiioonn CCoonnttrrooll 2200
- 21. MMaannyy TTCCPP ‘‘ffllaavvoorrss’’ • TCP New Reno • TCP SACK – requires sender and receiver both to support TCP SACK – possible state machine is complex. • TCP Vegas – adjusts window size based on difference between expected and actual RTT. • TCP Cubic CCoommppuutteerr NNeettwwoorrkkss:: TTCCPP CCoonnggeessttiioonn CCoonnttrrooll 2211
- 22. FFiigguurree 55..66 TThhrreeee--wwaayy TTCCPP HHaannddsshhaakkee CCoommppuutteerr NNeettwwoorrkkss:: TTCCPP CCoonnggeessttiioonn CCoonnttrrooll 2222
- 23. AAddaappttiivvee RReettrraannssmmiissssiioonnss RTT:: Round Trip Time between a pair of hosts on the Internet. • How to set the TimeOut value (RTO)? – The timeout value is set as a function of the expected RTT. – Consequences of a bad choice? Computer Networks: TTCCPP CCoonnggeessttiioonn CCoonnttrrooll 2233
- 24. OOrriiggiinnaall AAllggoorriitthhmm • Keep a running average of RTT and compute TimeOut as a function of this RTT. – Send packet and keep timestamp ts . – When ACK arrives, record timestamp ta . SampleRTT = ta - ts Computer Networks: TTCCPP CCoonnggeessttiioonn CCoonnttrrooll 2244
- 25. OOrriiggiinnaall AAllggoorriitthhmm Compute a weighted average: EEssttiimmaatteeddRRTTTT == αα xx EEssttiimmaatteeddRRTTTT ++ ((11-- αα)) xx SSaammpplleeRRTTTT Original TCP spec: αα iinn rraannggee ((00..88,,00..99)) TTiimmeeOOuutt == 22 xx EEssttiimmaatteeddRRTTTT Computer Networks: TTCCPP CCoonnggeessttiioonn CCoonnttrrooll 2255
- 26. KKaarrnn//PPaarrttiiddggee AAllggoorriitthhmm An obvious flaw in the original algorithm: Whenever there is a retransmission it is impossible to know whether to associate the ACK with the original packet or the retransmitted packet. Computer Networks: TTCCPP CCoonnggeessttiioonn CCoonnttrrooll 2266
- 27. FFiigguurree 55..1100 AAssssoocciiaattiinngg tthhee AACCKK?? Sender Receiver Original transmission Retransmission ACK Sender Receiver Original transmission ACK Retransmission (a) (b) Computer Networks: TTCCPP CCoonnggeessttiioonn CCoonnttrrooll 2277
- 28. KKaarrnn//PPaarrttiiddggee AAllggoorriitthhmm 1. Do not measure SSaammpplleeRRTTTT when sending packet more than once. 2. For each retransmission, set TTiimmeeOOuutt to double the last TTiimmeeOOuutt. { Note – this is a form of exponential backoff based on the believe that the lost packet is due to congestion.} Computer Networks: TTCCPP CCoonnggeessttiioonn CCoonnttrrooll 2288
- 29. JJaaccoobbssoonn//KKaarreellss AAllggoorriitthhmm The problem with the original algorithm is that it did not take into account the variance of SampleRTT. DDiiffffeerreennccee == SSaammpplleeRRTTTT –– EEssttiimmaatteeddRRTTTT EEssttiimmaatteeddRRTTTT == EEssttiimmaatteeddRRTTTT ++ ((δδ xx DDiiffffeerreennccee)) DDeevviiaattiioonn == δδ ((||DDiiffffeerreennccee|| - DDeevviiaattiioonn)) where δδ is a fraction between 0 and 1. Computer Networks: TTCCPP CCoonnggeessttiioonn CCoonnttrrooll 2299
- 30. JJaaccoobbssoonn//KKaarreellss AAllggoorriitthhmm TCP computes timeout using both the mean and variance of RTT TTiimmeeOOuutt == μμ xx EEssttiimmaatteeddRRTTTT ++ ΦΦ xx DDeevviiaattiioonn where based on experience μμ == 11 and ΦΦ == 44. Computer Networks: TTCCPP CCoonnggeessttiioonn CCoonnttrrooll 3300
- 31. TTCCPP CCoonnggeessttiioonn CCoonnttrrooll SSuummmmaarryy • TCP interacts with routers in the subnet and reacts to implicit congestion notification (packet drop) by reducing the TCP sender’s congestion window. • TCP increases congestion window using slow start or congestion avoidance. • Currently, the two most common versions of TCP are New Reno and Cubic CCoommppuutteerr NNeettwwoorrkkss:: TTCCPP CCoonnggeessttiioonn CCoonnttrrooll 3311
- 32. TTCCPP NNeeww RReennoo • Two problem scenarios with TCP Reno – bursty losses, Reno cannot recover from bursts of 3+ losses – Packets arriving out-of-order can yield duplicate acks when in fact there is no loss. • New Reno solution – try to determine the end of a burst loss. CCoommppuutteerr NNeettwwoorrkkss:: TTCCPP CCoonnggeessttiioonn CCoonnttrrooll 3322
- 33. TTCCPP NNeeww RReennoo • When duplicate ACKs trigger a retransmission for a lost packet, remember the highest packet sent from window in recover. • Upon receiving an ACK, – if ACK < recover => partial ACK – If ACK ≥ recover => new ACK CCoommppuutteerr NNeettwwoorrkkss:: TTCCPP CCoonnggeessttiioonn CCoonnttrrooll 3333
- 34. TTCCPP NNeeww RReennoo • Partial ACK implies another lost packet: retransmit next packet, inflate window and stay in fast recovery. • New ACK implies fast recovery is over: starting from 0.5 x cwnd proceed with congestion avoidance (linear increase). • New Reno recovers from n losses in n round trips. CCoommppuutteerr NNeettwwoorrkkss:: TTCCPP CCoonnggeessttiioonn CCoonnttrrooll 3344