tcp ex machina computer generated congestion control
play

TCP ex Machina: Computer-Generated Congestion Control Keith - PowerPoint PPT Presentation

Oct 16, 2022 •188 likes •1.67k views

TCP ex Machina: Computer-Generated Congestion Control Keith Winstein and Hari Balakrishnan MIT Computer Science and Artificial Intelligence Laboratory http://web.mit.edu/remy August 14, 2013 Introduction The problem Remy Evaluation

  1. Introduction The problem Remy Evaluation Discussion Dumbbell network Round-trip time Sender Receiver Queue Link ? ? Sender 2 Receiver 2 Keith Winstein and Hari Balakrishnan MIT Computer Science and Artificial Intelligence Laboratory TCP ex Machina: Computer-Generated Congestion Control

  2. Introduction The problem Remy Evaluation Discussion Dumbbell network Round-trip time Sender Receiver Queue Link Sender 2 Receiver 2 ... ... Sender n Receiver n Keith Winstein and Hari Balakrishnan MIT Computer Science and Artificial Intelligence Laboratory TCP ex Machina: Computer-Generated Congestion Control

  3. Introduction The problem Remy Evaluation Discussion Superrational congestion control At this moment, * do I: ◮ send a packet ◮ not send a packet? * Assuming every node is running the same algorithm. Keith Winstein and Hari Balakrishnan MIT Computer Science and Artificial Intelligence Laboratory TCP ex Machina: Computer-Generated Congestion Control

  4. Introduction The problem Remy Evaluation Discussion Remy: search for superrationality ◮ Remy searches for the best congestion-control algorithm ◮ Optimizes expected objective over prior assumptions ◮ Makes tractable by limiting available state Keith Winstein and Hari Balakrishnan MIT Computer Science and Artificial Intelligence Laboratory TCP ex Machina: Computer-Generated Congestion Control

  5. Introduction The problem Remy Evaluation Discussion A RemyCC tracks three congestion signals r ewma : moving average of interval between acks s ewma : . . . between sender timestamps echoed in acks rtt ratio : ratio of last RTT to smallest RTT so far Keith Winstein and Hari Balakrishnan MIT Computer Science and Artificial Intelligence Laboratory TCP ex Machina: Computer-Generated Congestion Control

  6. Introduction The problem Remy Evaluation Discussion Why these three congestion signals? ◮ Benefit can be measured empirically ◮ In our experiments, little help from adding more ◮ Other networks might find differently ◮ More signals increase search time Keith Winstein and Hari Balakrishnan MIT Computer Science and Artificial Intelligence Laboratory TCP ex Machina: Computer-Generated Congestion Control

  7. Introduction The problem Remy Evaluation Discussion A RemyCC maps each state to an action Rule ( r ewma , s ewma , rtt ratio ) → � m , b , τ � m Multiple to congestion window b Increment to congestion window Minimum interval between two outgoing packets τ Keith Winstein and Hari Balakrishnan MIT Computer Science and Artificial Intelligence Laboratory TCP ex Machina: Computer-Generated Congestion Control

  8. Introduction The problem Remy Evaluation Discussion Runtime for a RemyCC On ack: ◮ � m , b , τ � ← Rule ( r ewma , s ewma , rtt ratio ) ◮ cwnd ← m · cwnd + b Send packet if: ◮ cwnd > FlightSize , and ◮ last packet sent > τ ago Keith Winstein and Hari Balakrishnan MIT Computer Science and Artificial Intelligence Laboratory TCP ex Machina: Computer-Generated Congestion Control

  9. Introduction The problem Remy Evaluation Discussion Remy’s job Find piecewise-continuous Rule () that optimizes expected value of objective function. Keith Winstein and Hari Balakrishnan MIT Computer Science and Artificial Intelligence Laboratory TCP ex Machina: Computer-Generated Congestion Control

  10. Introduction The problem Remy Evaluation Discussion Remy example: 2D state space On ack: ◮ � m , b , τ � ← Rule ( r ewma , s ewma , rtt ratio ) Keith Winstein and Hari Balakrishnan MIT Computer Science and Artificial Intelligence Laboratory TCP ex Machina: Computer-Generated Congestion Control

  11. Introduction The problem Remy Evaluation Discussion Remy example: 2D state space On ack: ◮ � m , b , τ � ← Rule ( r ewma , s ewma , rtt ratio ) Keith Winstein and Hari Balakrishnan MIT Computer Science and Artificial Intelligence Laboratory TCP ex Machina: Computer-Generated Congestion Control

  12. Introduction The problem Remy Evaluation Discussion Remy example: Prior assumptions Quantity Distribution Units Link speed Uniform(10, 20) Mbps RTT Uniform(100, 200) ms Uniform(1, 16) n “On” process exp [ µ = 5] seconds “Off” process same Keith Winstein and Hari Balakrishnan MIT Computer Science and Artificial Intelligence Laboratory TCP ex Machina: Computer-Generated Congestion Control

  13. Introduction The problem Remy Evaluation Discussion Remy example: Objective � throughput i � � log delay i i Keith Winstein and Hari Balakrishnan MIT Computer Science and Artificial Intelligence Laboratory TCP ex Machina: Computer-Generated Congestion Control

  14. Introduction The problem Remy Evaluation Discussion One action for all states. Find the best value. r_ewma <?,?,?> s_ewma Keith Winstein and Hari Balakrishnan MIT Computer Science and Artificial Intelligence Laboratory TCP ex Machina: Computer-Generated Congestion Control

  15. Introduction The problem Remy Evaluation Discussion The best (single) action. Now split it on median. r_ewma <0.90,4,3.3> s_ewma Keith Winstein and Hari Balakrishnan MIT Computer Science and Artificial Intelligence Laboratory TCP ex Machina: Computer-Generated Congestion Control

  16. Introduction The problem Remy Evaluation Discussion Simulate <0.90,4,3.3> <0.90,4,3.3> r_ewma <0.90,4,3.3> <0.90,4,3.3> s_ewma Keith Winstein and Hari Balakrishnan MIT Computer Science and Artificial Intelligence Laboratory TCP ex Machina: Computer-Generated Congestion Control

  17. Introduction The problem Remy Evaluation Discussion Optimize each of the new actions <0.90,4,3.3> <0.90,4,3.3> r_ewma <0.90,4,3.3> <0.90,4,3.3> s_ewma Keith Winstein and Hari Balakrishnan MIT Computer Science and Artificial Intelligence Laboratory TCP ex Machina: Computer-Generated Congestion Control

  18. Introduction The problem Remy Evaluation Discussion Now split the most-used rule <0.70,6,53.5> <0.90,5,2.8> r_ewma <0.80,5,4.1> <0.60,19,76.2> s_ewma Keith Winstein and Hari Balakrishnan MIT Computer Science and Artificial Intelligence Laboratory TCP ex Machina: Computer-Generated Congestion Control

  19. Introduction The problem Remy Evaluation Discussion Simulate <0.70,6,53.5> <0.90,5,2.8> r_ewma <0.80,5,4.1> <0.80,5,4.1> <0.60,19,76.2> <0.80,5,4.1> <0.80,5,4.1> s_ewma Keith Winstein and Hari Balakrishnan MIT Computer Science and Artificial Intelligence Laboratory TCP ex Machina: Computer-Generated Congestion Control

  20. Introduction The problem Remy Evaluation Discussion Optimize <0.70,6,53.5> <0.90,5,2.8> r_ewma <0.80,5,4.1> <0.80,5,4.1> <0.60,19,76.2> <0.80,5,4.1> <0.80,5,4.1> s_ewma Keith Winstein and Hari Balakrishnan MIT Computer Science and Artificial Intelligence Laboratory TCP ex Machina: Computer-Generated Congestion Control

  21. Introduction The problem Remy Evaluation Discussion Split <0.60,17,13.3> <0.90,5,2.8> r_ewma <0.80,17,4.6> <0.80,8,3.3> <0.30,29,49.7> <0.80,7,16.9> <0.80,8,62.7> s_ewma Keith Winstein and Hari Balakrishnan MIT Computer Science and Artificial Intelligence Laboratory TCP ex Machina: Computer-Generated Congestion Control

  22. Introduction The problem Remy Evaluation Discussion Simulate <0.90,5,2.8> <0.60,17,13.3> <0.90,5,2.8> r_ewma <0.90,5,2.8> <0.90,5,2.8> <0.80,17,4.6> <0.80,8,3.3> <0.30,29,49.7> <0.80,7,16.9> <0.80,8,62.7> s_ewma Keith Winstein and Hari Balakrishnan MIT Computer Science and Artificial Intelligence Laboratory TCP ex Machina: Computer-Generated Congestion Control

  23. Introduction The problem Remy Evaluation Discussion Optimize <0.90,5,2.8> <0.60,17,13.3> <0.90,5,2.8> r_ewma <0.90,5,2.8> <0.90,5,2.8> <0.80,17,4.6> <0.80,8,3.3> <0.30,29,49.7> <0.80,7,16.9> <0.80,8,62.7> s_ewma Keith Winstein and Hari Balakrishnan MIT Computer Science and Artificial Intelligence Laboratory TCP ex Machina: Computer-Generated Congestion Control

  24. Introduction The problem Remy Evaluation Discussion Split <1.30,213,8.9> <0.80,30,6.8> <1.00,-139,9.1> r_ewma <0.90,-84,9.2> <1.40,32,3.3> <0.90,-31,2.8> <0.50,84,3.4> <0.60,40,42.4> <1.30,184,15.3> <0.60,8,57.2> s_ewma Keith Winstein and Hari Balakrishnan MIT Computer Science and Artificial Intelligence Laboratory TCP ex Machina: Computer-Generated Congestion Control

  25. Introduction The problem Remy Evaluation Discussion Simulate <1.30,213,8.9> <0.80,30,6.8> <1.00,-139,9.1> r_ewma <0.90,-84,9.2> <1.40,32,3.3> <1.40,32,3.3> <1.40,32,3.3> <1.40,32,3.3> <0.90,-31,2.8> <0.50,84,3.4> <0.60,40,42.4> <1.30,184,15.3> <0.60,8,57.2> s_ewma Keith Winstein and Hari Balakrishnan MIT Computer Science and Artificial Intelligence Laboratory TCP ex Machina: Computer-Generated Congestion Control

  26. Introduction The problem Remy Evaluation Discussion Optimize <1.30,213,8.9> <0.80,30,6.8> <1.00,-139,9.1> r_ewma <0.90,-84,9.2> <1.40,32,3.3> <1.40,32,3.3> <1.40,32,3.3> <1.40,32,3.3> <0.90,-31,2.8> <0.50,84,3.4> <0.60,40,42.4> <1.30,184,15.3> <0.60,8,57.2> s_ewma Keith Winstein and Hari Balakrishnan MIT Computer Science and Artificial Intelligence Laboratory TCP ex Machina: Computer-Generated Congestion Control

  27. Introduction The problem Remy Evaluation Discussion Split <1.20,221,9.0> <1.00,206,6.6> <1.00,-220,2.8> r_ewma <0.90,-92,17.1> <1.20,170,3.0> <1.60,186,3.0> <1.40,185,4.9> <1.40,170,3.2> <1.00,-96,4.2> <0.50,216,3.5> <0.50,256,27.2> <0.90,201,16.2> <0.80,250,56.3> s_ewma Keith Winstein and Hari Balakrishnan MIT Computer Science and Artificial Intelligence Laboratory TCP ex Machina: Computer-Generated Congestion Control

  28. Introduction The problem Remy Evaluation Discussion Simulate <1.20,221,9.0> <1.00,206,6.6> <1.00,-220,2.8> r_ewma <0.90,-92,17.1> <1.20,170,3.0> <1.20,170,3.0> <1.60,186,3.0> <1.20,170,3.0> <1.20,170,3.0> <1.40,185,4.9> <1.40,170,3.2> <1.00,-96,4.2> <0.50,216,3.5> <0.50,256,27.2> <0.90,201,16.2> <0.80,250,56.3> s_ewma Keith Winstein and Hari Balakrishnan MIT Computer Science and Artificial Intelligence Laboratory TCP ex Machina: Computer-Generated Congestion Control

  29. Introduction The problem Remy Evaluation Discussion Optimize <1.20,221,9.0> <1.00,206,6.6> <1.00,-220,2.8> r_ewma <0.90,-92,17.1> <1.20,170,3.0> <1.20,170,3.0> <1.60,186,3.0> <1.20,170,3.0> <1.20,170,3.0> <1.40,185,4.9> <1.40,170,3.2> <1.00,-96,4.2> <0.50,216,3.5> <0.50,256,27.2> <0.90,201,16.2> <0.80,250,56.3> s_ewma Keith Winstein and Hari Balakrishnan MIT Computer Science and Artificial Intelligence Laboratory TCP ex Machina: Computer-Generated Congestion Control

  30. Introduction The problem Remy Evaluation Discussion Split <1.90,253,8.1> <1.20,232,6.6> <1.00,-254,3.8> r_ewma <0.70,-157,70.0> <1.30,251,2.8> <1.10,250,2.9> <1.80,255,2.8> <1.00,254,4.6> <1.40,187,2.9> <1.50,194,4.1> <1.50,251,3.1> <0.80,-105,4.3> <0.60,234,3.3> <0.60,256,28.0> <0.80,203,16.3> <0.80,253,71.4> s_ewma Keith Winstein and Hari Balakrishnan MIT Computer Science and Artificial Intelligence Laboratory TCP ex Machina: Computer-Generated Congestion Control

  31. Introduction The problem Remy Evaluation Discussion Simulate <1.90,253,8.1> <1.20,232,6.6> <1.00,-254,3.8> r_ewma <1.30,251,2.8> <1.30,251,2.8> <0.70,-157,70.0> <1.10,250,2.9> <1.30,251,2.8> <1.30,251,2.8> <1.80,255,2.8> <1.00,254,4.6> <1.40,187,2.9> <1.50,194,4.1> <1.50,251,3.1> <0.80,-105,4.3> <0.60,234,3.3> <0.60,256,28.0> <0.80,203,16.3> <0.80,253,71.4> s_ewma Keith Winstein and Hari Balakrishnan MIT Computer Science and Artificial Intelligence Laboratory TCP ex Machina: Computer-Generated Congestion Control

  32. Introduction The problem Remy Evaluation Discussion Optimize <1.90,253,8.1> <1.20,232,6.6> <1.00,-254,3.8> r_ewma <1.30,251,2.8> <1.30,251,2.8> <0.70,-157,70.0> <1.10,250,2.9> <1.30,251,2.8> <1.30,251,2.8> <1.80,255,2.8> <1.00,254,4.6> <1.40,187,2.9> <1.50,194,4.1> <1.50,251,3.1> <0.80,-105,4.3> <0.60,234,3.3> <0.60,256,28.0> <0.80,203,16.3> <0.80,253,71.4> s_ewma Keith Winstein and Hari Balakrishnan MIT Computer Science and Artificial Intelligence Laboratory TCP ex Machina: Computer-Generated Congestion Control

  33. Introduction The problem Remy Evaluation Discussion Split <1.70,256,8.1> <1.20,248,6.5> <1.10,-255,4.7> r_ewma <1.00,251,3.3> <1.30,253,3.6> <0.80,-222,69.3> <1.40,252,2.9> <1.10,251,2.6> <1.10,255,2.8> <2.00,256,2.7> <0.80,255,4.7> <1.10,253,2.9> <1.20,237,4.0> <1.50,254,2.9> <0.80,-243,3.6> <0.70,252,3.2> <0.50,256,27.3> <0.90,227,15.5> <1.20,255,71.1> s_ewma Keith Winstein and Hari Balakrishnan MIT Computer Science and Artificial Intelligence Laboratory TCP ex Machina: Computer-Generated Congestion Control

  34. Introduction The problem Remy Evaluation Discussion Simulate <1.70,256,8.1> <1.20,248,6.5> <1.10,-255,4.7> r_ewma <1.00,251,3.3> <1.30,253,3.6> <0.80,-222,69.3> <1.40,252,2.9> <1.10,251,2.6> <1.10,255,2.8> <2.00,256,2.7> <0.80,255,4.7> <1.10,253,2.9> <1.20,237,4.0> <1.50,254,2.9> <0.80,-243,3.6> <0.70,252,3.2> <0.50,256,27.3> <0.90,227,15.5> <1.20,255,71.1> <0.90,227,15.5> <0.90,227,15.5> <0.90,227,15.5> s_ewma Keith Winstein and Hari Balakrishnan MIT Computer Science and Artificial Intelligence Laboratory TCP ex Machina: Computer-Generated Congestion Control

  35. Introduction The problem Remy Evaluation Discussion Optimize <1.70,256,8.1> <1.20,248,6.5> <1.10,-255,4.7> r_ewma <1.00,251,3.3> <1.30,253,3.6> <0.80,-222,69.3> <1.40,252,2.9> <1.10,251,2.6> <1.10,255,2.8> <2.00,256,2.7> <0.80,255,4.7> <1.10,253,2.9> <1.20,237,4.0> <1.50,254,2.9> <0.80,-243,3.6> <0.70,252,3.2> <0.50,256,27.3> <0.90,227,15.5> <1.20,255,71.1> <0.90,227,15.5> <0.90,227,15.5> <0.90,227,15.5> s_ewma Keith Winstein and Hari Balakrishnan MIT Computer Science and Artificial Intelligence Laboratory TCP ex Machina: Computer-Generated Congestion Control

  36. Introduction The problem Remy Evaluation Discussion Split <1.40,256,8.1> <0.80,256,6.6> <1.00,-256,5.6> r_ewma <0.80,251,3.0> <1.00,253,2.7> <0.70,-222,75.8> <1.10,256,2.9> <1.10,251,3.0> <1.10,256,2.9> <2.00,256,3.2> <0.80,255,4.0> <1.00,254,3.0> <1.00,253,3.8> <1.60,256,3.1> <1.00,-253,2.8> <0.50,253,3.7> <0.60,256,23.5> <1.20,244,13.8> <1.40,256,65.5> <1.00,227,66.8> <0.80,227,37.3> <1.00,227,13.8> s_ewma Keith Winstein and Hari Balakrishnan MIT Computer Science and Artificial Intelligence Laboratory TCP ex Machina: Computer-Generated Congestion Control

  37. Introduction The problem Remy Evaluation Discussion Simulate <1.40,256,8.1> <0.80,256,6.6> <1.00,-256,5.6> r_ewma <0.80,251,3.0> <1.00,253,2.7> <0.70,-222,75.8> <1.10,256,2.9> <1.10,251,3.0> <1.10,256,2.9> <2.00,256,3.2> <1.00,254,3.0> <1.00,254,3.0> <0.80,255,4.0> <1.00,254,3.0> <1.00,254,3.0> <1.00,253,3.8> <1.60,256,3.1> <1.00,-253,2.8> <0.50,253,3.7> <0.60,256,23.5> <1.20,244,13.8> <1.40,256,65.5> <1.00,227,66.8> <0.80,227,37.3> <1.00,227,13.8> s_ewma Keith Winstein and Hari Balakrishnan MIT Computer Science and Artificial Intelligence Laboratory TCP ex Machina: Computer-Generated Congestion Control

  38. Introduction The problem Remy Evaluation Discussion Optimize <1.40,256,8.1> <0.80,256,6.6> <1.00,-256,5.6> r_ewma <0.80,251,3.0> <1.00,253,2.7> <0.70,-222,75.8> <1.10,256,2.9> <1.10,251,3.0> <1.10,256,2.9> <2.00,256,3.2> <1.00,254,3.0> <1.00,254,3.0> <0.80,255,4.0> <1.00,254,3.0> <1.00,254,3.0> <1.00,253,3.8> <1.60,256,3.1> <1.00,-253,2.8> <0.50,253,3.7> <0.60,256,23.5> <1.20,244,13.8> <1.40,256,65.5> <1.00,227,66.8> <0.80,227,37.3> <1.00,227,13.8> s_ewma Keith Winstein and Hari Balakrishnan MIT Computer Science and Artificial Intelligence Laboratory TCP ex Machina: Computer-Generated Congestion Control

  39. Introduction The problem Remy Evaluation Discussion Split <1.80,256,8.7> <0.50,256,5.8> <1.00,-256,5.6> r_ewma <0.80,251,4.0> <1.30,253,3.7> <0.80,-230,19.1> <0.90,256,2.9> <1.40,251,2.2> <1.30,256,2.9> <1.80,256,3.9> <1.40,254,3.0> <1.10,254,4.7> <1.10,256,4.5> <1.10,254,2.1> <1.00,254,2.8> <1.10,253,3.8> <1.40,256,2.9> <1.10,-256,3.6> <0.70,254,3.7> <0.80,256,23.4> <1.10,254,13.8> <1.90,256,64.0> <0.80,227,7.7> <0.90,227,45.2> <1.10,227,25.0> s_ewma Keith Winstein and Hari Balakrishnan MIT Computer Science and Artificial Intelligence Laboratory TCP ex Machina: Computer-Generated Congestion Control

  40. Introduction The problem Remy Evaluation Discussion Simulate <1.80,256,8.7> <0.50,256,5.8> <1.00,-256,5.6> r_ewma <0.80,251,4.0> <1.30,253,3.7> <0.80,-230,19.1> <0.90,256,2.9> <1.40,251,2.2> <1.30,256,2.9> <1.80,256,3.9> <1.40,254,3.0> <1.10,254,4.7> <1.10,256,4.5> <1.10,254,2.1> <1.00,254,2.8> <1.10,253,3.8> <1.40,256,2.9> <0.70,254,3.7> <0.70,254,3.7> <1.10,-256,3.6> <0.70,254,3.7> <0.70,254,3.7> <0.80,256,23.4> <1.10,254,13.8> <1.90,256,64.0> <0.80,227,7.7> <0.90,227,45.2> <1.10,227,25.0> s_ewma Keith Winstein and Hari Balakrishnan MIT Computer Science and Artificial Intelligence Laboratory TCP ex Machina: Computer-Generated Congestion Control

  41. Introduction The problem Remy Evaluation Discussion Optimize <1.80,256,8.7> <0.50,256,5.8> <1.00,-256,5.6> r_ewma <0.80,251,4.0> <1.30,253,3.7> <0.80,-230,19.1> <0.90,256,2.9> <1.40,251,2.2> <1.30,256,2.9> <1.80,256,3.9> <1.40,254,3.0> <1.10,254,4.7> <1.10,256,4.5> <1.10,254,2.1> <1.00,254,2.8> <1.10,253,3.8> <1.40,256,2.9> <0.70,254,3.7> <0.70,254,3.7> <1.10,-256,3.6> <0.70,254,3.7> <0.70,254,3.7> <0.80,256,23.4> <1.10,254,13.8> <1.90,256,64.0> <0.80,227,7.7> <0.90,227,45.2> <1.10,227,25.0> s_ewma Keith Winstein and Hari Balakrishnan MIT Computer Science and Artificial Intelligence Laboratory TCP ex Machina: Computer-Generated Congestion Control

  42. Introduction The problem Remy Evaluation Discussion Split <1.90,256,8.6> <0.40,256,5.3> <1.10,-256,6.7> r_ewma <0.90,253,3.1> <1.30,253,2.9> <0.80,-230,24.8> <1.10,256,2.1> <1.50,251,3.1> <0.90,256,2.9> <1.80,256,3.0> <1.20,256,2.8> <1.30,254,6.3> <0.90,256,4.1> <1.30,255,3.6> <1.00,256,2.9> <1.30,254,3.9> <1.20,256,3.1> <1.00,255,3.8> <0.80,254,3.8> <1.40,-256,4.2> <0.70,255,3.7> <0.50,256,16.4> <0.80,256,17.0> <0.90,254,13.1> <1.80,256,50.6> <0.80,227,7.1> <0.90,227,52.4> <1.00,227,32.0> s_ewma Keith Winstein and Hari Balakrishnan MIT Computer Science and Artificial Intelligence Laboratory TCP ex Machina: Computer-Generated Congestion Control

  43. Introduction The problem Remy Evaluation Discussion Simulate <1.90,256,8.6> <0.40,256,5.3> <1.10,-256,6.7> r_ewma <0.90,253,3.1> <1.30,253,2.9> <0.80,-230,24.8> <1.10,256,2.1> <1.50,251,3.1> <0.90,256,2.9> <1.80,256,3.0> <1.20,256,2.8> <1.30,254,6.3> <0.90,256,4.1> <1.30,255,3.6> <1.00,256,2.9> <1.20,256,3.1> <1.20,256,3.1> <1.30,254,3.9> <1.20,256,3.1> <1.20,256,3.1> <1.00,255,3.8> <0.80,254,3.8> <1.40,-256,4.2> <0.70,255,3.7> <0.50,256,16.4> <0.80,256,17.0> <0.90,254,13.1> <1.80,256,50.6> <0.80,227,7.1> <0.90,227,52.4> <1.00,227,32.0> s_ewma Keith Winstein and Hari Balakrishnan MIT Computer Science and Artificial Intelligence Laboratory TCP ex Machina: Computer-Generated Congestion Control

  44. Introduction The problem Remy Evaluation Discussion Optimize <1.90,256,8.6> <0.40,256,5.3> <1.10,-256,6.7> r_ewma <0.90,253,3.1> <1.30,253,2.9> <0.80,-230,24.8> <1.10,256,2.1> <1.50,251,3.1> <0.90,256,2.9> <1.80,256,3.0> <1.20,256,2.8> <1.30,254,6.3> <0.90,256,4.1> <1.30,255,3.6> <1.00,256,2.9> <1.20,256,3.1> <1.20,256,3.1> <1.30,254,3.9> <1.20,256,3.1> <1.20,256,3.1> <1.00,255,3.8> <0.80,254,3.8> <1.40,-256,4.2> <0.70,255,3.7> <0.50,256,16.4> <0.80,256,17.0> <0.90,254,13.1> <1.80,256,50.6> <0.80,227,7.1> <0.90,227,52.4> <1.00,227,32.0> s_ewma Keith Winstein and Hari Balakrishnan MIT Computer Science and Artificial Intelligence Laboratory TCP ex Machina: Computer-Generated Congestion Control

  45. Introduction The problem Remy Evaluation Discussion Split <1.30,256,8.5> <0.30,256,7.0> <1.00,-256,5.2> r_ewma <0.90,253,3.0> <1.30,253,3.1> <0.40,-247,11.2> <1.50,256,2.9> <1.60,251,3.1> <0.90,256,2.8> <1.70,256,2.2> <1.30,256,2.9> <1.20,254,3.9> <0.70,256,4.0> <1.50,255,4.2> <1.00,256,2.8> <1.30,256,3.1> <1.10,256,4.9> <1.50,254,4.1> <1.10,256,3.8> <0.90,256,3.1> <0.80,256,9.1> <0.90,256,3.1> <1.20,-256,3.2> <0.50,256,3.8> <0.60,256,28.5> <0.70,256,10.6> <0.70,256,13.4> <1.60,256,44.1> <1.00,227,14.3> <0.90,227,57.9> <1.00,227,24.0> s_ewma Keith Winstein and Hari Balakrishnan MIT Computer Science and Artificial Intelligence Laboratory TCP ex Machina: Computer-Generated Congestion Control

  46. Introduction The problem Remy Evaluation Discussion Simulate <1.30,256,8.5> <0.30,256,7.0> <1.00,-256,5.2> r_ewma <0.90,253,3.0> <1.30,253,3.1> <0.40,-247,11.2> <1.50,256,2.9> <1.60,251,3.1> <0.90,256,2.8> <1.70,256,2.2> <1.30,256,2.9> <1.20,254,3.9> <0.70,256,4.0> <1.50,255,4.2> <1.00,256,2.8> <1.30,256,3.1> <1.10,256,4.9> <1.50,254,4.1> <1.10,256,3.8> <0.90,256,3.1> <0.90,256,3.1> <0.90,256,3.1> <0.80,256,9.1> <0.90,256,3.1> <0.90,256,3.1> <1.20,-256,3.2> <0.50,256,3.8> <0.60,256,28.5> <0.70,256,10.6> <0.70,256,13.4> <1.60,256,44.1> <1.00,227,14.3> <0.90,227,57.9> <1.00,227,24.0> s_ewma Keith Winstein and Hari Balakrishnan MIT Computer Science and Artificial Intelligence Laboratory TCP ex Machina: Computer-Generated Congestion Control

  47. Introduction The problem Remy Evaluation Discussion Optimize <1.30,256,8.5> <0.30,256,7.0> <1.00,-256,5.2> r_ewma <0.90,253,3.0> <1.30,253,3.1> <0.40,-247,11.2> <1.50,256,2.9> <1.60,251,3.1> <0.90,256,2.8> <1.70,256,2.2> <1.30,256,2.9> <1.20,254,3.9> <0.70,256,4.0> <1.50,255,4.2> <1.00,256,2.8> <1.30,256,3.1> <1.10,256,4.9> <1.50,254,4.1> <1.10,256,3.8> <0.90,256,3.1> <0.90,256,3.1> <0.90,256,3.1> <0.80,256,9.1> <0.90,256,3.1> <0.90,256,3.1> <1.20,-256,3.2> <0.50,256,3.8> <0.60,256,28.5> <0.70,256,10.6> <0.70,256,13.4> <1.60,256,44.1> <1.00,227,14.3> <0.90,227,57.9> <1.00,227,24.0> s_ewma Keith Winstein and Hari Balakrishnan MIT Computer Science and Artificial Intelligence Laboratory TCP ex Machina: Computer-Generated Congestion Control

  48. Introduction The problem Remy Evaluation Discussion Split <1.30,256,7.8> <0.10,256,3.7> <1.20,-256,5.2> r_ewma <0.50,255,3.0> <1.50,254,3.1> <0.20,-256,5.7> <1.50,256,3.2> <1.50,251,9.0> <1.10,256,2.8> <1.60,256,2.8> <1.10,256,3.1> <1.60,254,3.2> <0.70,256,3.9> <1.90,255,3.2> <0.50,256,2.9> <1.20,256,2.9> <1.50,256,6.3> <1.60,254,4.0> <0.90,256,4.2> <0.70,256,2.7> <0.80,256,3.9> <0.60,256,3.1> <0.90,256,8.3> <1.20,256,3.9> <1.00,256,3.7> <1.10,-256,3.1> <0.50,256,4.0> <0.30,256,32.3> <0.70,256,10.5> <0.70,256,12.4> <1.50,256,44.0> <1.10,227,14.9> <0.80,227,50.9> <1.00,227,22.4> s_ewma Keith Winstein and Hari Balakrishnan MIT Computer Science and Artificial Intelligence Laboratory TCP ex Machina: Computer-Generated Congestion Control

  49. Introduction The problem Remy Evaluation Discussion Simulate <1.30,256,7.8> <0.10,256,3.7> <1.20,-256,5.2> r_ewma <0.50,255,3.0> <1.50,254,3.1> <0.20,-256,5.7> <1.50,256,3.2> <1.50,251,9.0> <1.10,256,2.8> <1.60,256,2.8> <1.10,256,3.1> <1.60,254,3.2> <0.70,256,3.9> <1.90,255,3.2> <0.50,256,2.9> <1.20,256,2.9> <1.50,256,6.3> <1.60,254,4.0> <0.90,256,4.2> <0.70,256,2.7> <0.80,256,3.9> <0.60,256,3.1> <0.90,256,8.3> <1.20,256,3.9> <1.00,256,3.7> <1.10,-256,3.1> <0.50,256,4.0> <0.30,256,32.3> <0.70,256,10.5> <0.70,256,12.4> <0.70,256,12.4> <1.50,256,44.0> <1.10,227,14.9> <0.70,256,12.4> <0.70,256,12.4> <0.80,227,50.9> <1.00,227,22.4> s_ewma Keith Winstein and Hari Balakrishnan MIT Computer Science and Artificial Intelligence Laboratory TCP ex Machina: Computer-Generated Congestion Control

  50. Introduction The problem Remy Evaluation Discussion Optimize <1.30,256,7.8> <0.10,256,3.7> <1.20,-256,5.2> r_ewma <0.50,255,3.0> <1.50,254,3.1> <0.20,-256,5.7> <1.50,256,3.2> <1.50,251,9.0> <1.10,256,2.8> <1.60,256,2.8> <1.10,256,3.1> <1.60,254,3.2> <0.70,256,3.9> <1.90,255,3.2> <0.50,256,2.9> <1.20,256,2.9> <1.50,256,6.3> <1.60,254,4.0> <0.90,256,4.2> <0.70,256,2.7> <0.80,256,3.9> <0.60,256,3.1> <0.90,256,8.3> <1.20,256,3.9> <1.00,256,3.7> <1.10,-256,3.1> <0.50,256,4.0> <0.30,256,32.3> <0.70,256,10.5> <0.70,256,12.4> <0.70,256,12.4> <1.50,256,44.0> <1.10,227,14.9> <0.70,256,12.4> <0.70,256,12.4> <0.80,227,50.9> <1.00,227,22.4> s_ewma Keith Winstein and Hari Balakrishnan MIT Computer Science and Artificial Intelligence Laboratory TCP ex Machina: Computer-Generated Congestion Control

  51. Introduction The problem Remy Evaluation Discussion Split <0.90,256,7.9> <0.30,256,4.4> <0.70,-256,11.7> r_ewma <0.50,255,4.6> <1.40,254,3.0> <0.30,-256,10.3> <1.30,256,3.1> <1.40,252,7.4> <1.10,256,3.6> <1.80,256,3.4> <0.70,256,2.9> <1.50,254,3.4> <0.50,256,2.4> <1.80,256,3.2> <0.60,256,2.9> <0.90,256,3.2> <1.30,256,5.6> <1.90,255,4.4> <0.80,256,4.0> <1.00,256,2.9> <1.00,256,3.2> <0.40,256,2.5> <0.60,256,9.2> <1.30,256,3.8> <1.20,256,4.4> <1.20,-256,3.7> <0.50,256,4.2> <0.70,256,26.5> <0.50,256,10.0> <0.40,256,10.8> <0.50,256,3.4> <1.30,256,32.7> <1.50,227,60.6> <0.50,256,26.2> <0.90,256,19.6> <0.80,227,57.3> <1.20,227,27.2> s_ewma Keith Winstein and Hari Balakrishnan MIT Computer Science and Artificial Intelligence Laboratory TCP ex Machina: Computer-Generated Congestion Control

  52. Introduction The problem Remy Evaluation Discussion Simulate <0.90,256,7.9> <0.30,256,4.4> <0.70,-256,11.7> r_ewma <0.50,255,4.6> <1.40,254,3.0> <0.30,-256,10.3> <1.30,256,3.1> <1.40,252,7.4> <1.10,256,3.6> <1.80,256,3.4> <0.70,256,2.9> <1.50,254,3.4> <0.50,256,2.4> <1.80,256,3.2> <0.60,256,2.9> <0.90,256,3.2> <1.30,256,5.6> <1.90,255,4.4> <0.80,256,4.0> <1.00,256,2.9> <1.00,256,3.2> <0.40,256,2.5> <0.60,256,9.2> <1.30,256,3.8> <1.20,256,4.4> <1.20,-256,3.7> <0.50,256,4.2> <0.50,256,4.2> <0.70,256,26.5> <0.50,256,4.2> <0.50,256,4.2> <0.50,256,10.0> <0.40,256,10.8> <0.50,256,3.4> <1.30,256,32.7> <1.50,227,60.6> <0.50,256,26.2> <0.90,256,19.6> <0.80,227,57.3> <1.20,227,27.2> s_ewma Keith Winstein and Hari Balakrishnan MIT Computer Science and Artificial Intelligence Laboratory TCP ex Machina: Computer-Generated Congestion Control

  53. Introduction The problem Remy Evaluation Discussion Optimize <0.90,256,7.9> <0.30,256,4.4> <0.70,-256,11.7> r_ewma <0.50,255,4.6> <1.40,254,3.0> <0.30,-256,10.3> <1.30,256,3.1> <1.40,252,7.4> <1.10,256,3.6> <1.80,256,3.4> <0.70,256,2.9> <1.50,254,3.4> <0.50,256,2.4> <1.80,256,3.2> <0.60,256,2.9> <0.90,256,3.2> <1.30,256,5.6> <1.90,255,4.4> <0.80,256,4.0> <1.00,256,2.9> <1.00,256,3.2> <0.40,256,2.5> <0.60,256,9.2> <1.30,256,3.8> <1.20,256,4.4> <1.20,-256,3.7> <0.50,256,4.2> <0.50,256,4.2> <0.70,256,26.5> <0.50,256,4.2> <0.50,256,4.2> <0.50,256,10.0> <0.40,256,10.8> <0.50,256,3.4> <1.30,256,32.7> <1.50,227,60.6> <0.50,256,26.2> <0.90,256,19.6> <0.80,227,57.3> <1.20,227,27.2> s_ewma Keith Winstein and Hari Balakrishnan MIT Computer Science and Artificial Intelligence Laboratory TCP ex Machina: Computer-Generated Congestion Control

  54. Introduction The problem Remy Evaluation Discussion Split <0.70,256,8.1> <0.70,256,3.6> <0.70,-256,17.1> r_ewma <0.50,255,11.8> <1.40,254,10.1> <0.30,-256,15.8> <1.20,256,3.9> <1.40,252,53.8> <1.00,256,5.4> <1.80,256,3.3> <0.20,256,3.1> <1.40,254,3.0> <0.10,256,2.3> <1.60,256,4.0> <0.70,256,3.1> <0.90,256,3.0> <0.70,256,3.8> <1.70,256,5.4> <1.40,256,4.1> <0.90,256,3.5> <0.90,256,3.2> <0.30,256,4.3> <0.40,256,8.4> <1.30,256,5.2> <1.40,256,4.4> <1.30,-256,4.3> <0.50,256,4.0> <0.40,256,4.5> <0.50,256,26.7> <0.50,256,5.2> <0.50,256,4.0> <0.70,256,9.2> <0.90,256,10.7> <0.90,256,4.2> <0.90,256,33.4> <1.30,228,65.6> <0.60,256,33.4> <0.90,256,12.3> <0.80,227,63.7> <1.10,227,39.1> s_ewma Keith Winstein and Hari Balakrishnan MIT Computer Science and Artificial Intelligence Laboratory TCP ex Machina: Computer-Generated Congestion Control

  55. Introduction The problem Remy Evaluation Discussion Simulate <0.70,256,8.1> <0.70,256,3.6> <0.70,-256,17.1> r_ewma <0.50,255,11.8> <1.40,254,10.1> <0.30,-256,15.8> <1.20,256,3.9> <1.40,252,53.8> <1.00,256,5.4> <1.80,256,3.3> <0.20,256,3.1> <1.40,254,3.0> <0.10,256,2.3> <1.60,256,4.0> <0.70,256,3.1> <0.90,256,3.0> <0.70,256,3.8> <1.70,256,5.4> <1.40,256,4.1> <0.90,256,3.5> <0.90,256,3.2> <0.30,256,4.3> <0.40,256,8.4> <1.30,256,5.2> <1.40,256,4.4> <1.30,-256,4.3> <0.50,256,4.0> <0.40,256,4.5> <0.50,256,26.7> <0.50,256,5.2> <0.50,256,4.0> <0.70,256,9.2> <0.90,256,10.7> <0.90,256,10.7> <0.90,256,4.2> <0.90,256,33.4> <1.30,228,65.6> <0.90,256,10.7> <0.90,256,10.7> <0.60,256,33.4> <0.90,256,12.3> <0.80,227,63.7> <1.10,227,39.1> s_ewma Keith Winstein and Hari Balakrishnan MIT Computer Science and Artificial Intelligence Laboratory TCP ex Machina: Computer-Generated Congestion Control

  56. Introduction The problem Remy Evaluation Discussion Optimize <0.70,256,8.1> <0.70,256,3.6> <0.70,-256,17.1> r_ewma <0.50,255,11.8> <1.40,254,10.1> <0.30,-256,15.8> <1.20,256,3.9> <1.40,252,53.8> <1.00,256,5.4> <1.80,256,3.3> <0.20,256,3.1> <1.40,254,3.0> <0.10,256,2.3> <1.60,256,4.0> <0.70,256,3.1> <0.90,256,3.0> <0.70,256,3.8> <1.70,256,5.4> <1.40,256,4.1> <0.90,256,3.5> <0.90,256,3.2> <0.30,256,4.3> <0.40,256,8.4> <1.30,256,5.2> <1.40,256,4.4> <1.30,-256,4.3> <0.50,256,4.0> <0.40,256,4.5> <0.50,256,26.7> <0.50,256,5.2> <0.50,256,4.0> <0.70,256,9.2> <0.90,256,10.7> <0.90,256,10.7> <0.90,256,4.2> <0.90,256,33.4> <1.30,228,65.6> <0.90,256,10.7> <0.90,256,10.7> <0.60,256,33.4> <0.90,256,12.3> <0.80,227,63.7> <1.10,227,39.1> s_ewma Keith Winstein and Hari Balakrishnan MIT Computer Science and Artificial Intelligence Laboratory TCP ex Machina: Computer-Generated Congestion Control

  57. Introduction The problem Remy Evaluation Discussion Split <0.80,256,9.7> <0.70,256,4.6> <0.90,-256,24.3> r_ewma <0.50,255,18.2> <1.40,254,22.9> <0.40,-256,60.0> <1.20,256,7.8> <1.50,252,4.2> <1.00,256,57.3> <1.90,256,3.1> <0.30,256,3.1> <1.60,254,3.0> <0.20,256,1.4> <1.50,256,7.9> <1.00,256,3.2> <0.80,256,3.1> <0.50,256,3.2> <1.90,256,3.9> <1.50,256,4.9> <0.70,256,3.5> <0.80,256,3.3> <0.70,256,3.5> <0.10,256,7.5> <1.20,256,5.0> <1.40,256,4.4> <1.40,-256,3.7> <0.50,256,3.8> <0.30,256,10.8> <0.40,256,21.3> <0.50,256,11.6> <0.50,256,4.0> <0.70,256,8.5> <1.20,256,9.1> <1.20,256,16.2> <1.00,256,5.0> <1.10,256,39.5> <1.40,228,28.2> <0.90,256,11.8> <0.70,256,7.3> <0.80,256,39.9> <1.00,256,17.9> <0.70,227,8.5> <1.10,227,56.0> s_ewma Keith Winstein and Hari Balakrishnan MIT Computer Science and Artificial Intelligence Laboratory TCP ex Machina: Computer-Generated Congestion Control

  58. Introduction The problem Remy Evaluation Discussion Simulate <0.80,256,9.7> <0.70,256,4.6> <0.90,-256,24.3> r_ewma <0.50,255,18.2> <1.40,254,22.9> <0.40,-256,60.0> <1.20,256,7.8> <1.50,252,4.2> <1.00,256,57.3> <1.90,256,3.1> <0.30,256,3.1> <1.60,254,3.0> <0.20,256,1.4> <1.00,256,3.2> <1.00,256,3.2> <1.50,256,7.9> <1.00,256,3.2> <1.00,256,3.2> <0.80,256,3.1> <0.50,256,3.2> <1.90,256,3.9> <1.50,256,4.9> <0.70,256,3.5> <0.80,256,3.3> <0.70,256,3.5> <0.10,256,7.5> <1.20,256,5.0> <1.40,256,4.4> <1.40,-256,3.7> <0.50,256,3.8> <0.30,256,10.8> <0.40,256,21.3> <0.50,256,11.6> <0.50,256,4.0> <0.70,256,8.5> <1.20,256,9.1> <1.20,256,16.2> <1.00,256,5.0> <1.10,256,39.5> <1.40,228,28.2> <0.90,256,11.8> <0.70,256,7.3> <0.80,256,39.9> <1.00,256,17.9> <0.70,227,8.5> <1.10,227,56.0> s_ewma Keith Winstein and Hari Balakrishnan MIT Computer Science and Artificial Intelligence Laboratory TCP ex Machina: Computer-Generated Congestion Control

  59. Introduction The problem Remy Evaluation Discussion Optimize <0.80,256,9.7> <0.70,256,4.6> <0.90,-256,24.3> r_ewma <0.50,255,18.2> <1.40,254,22.9> <0.40,-256,60.0> <1.20,256,7.8> <1.50,252,4.2> <1.00,256,57.3> <1.90,256,3.1> <0.30,256,3.1> <1.60,254,3.0> <0.20,256,1.4> <1.00,256,3.2> <1.00,256,3.2> <1.50,256,7.9> <1.00,256,3.2> <1.00,256,3.2> <0.80,256,3.1> <0.50,256,3.2> <1.90,256,3.9> <1.50,256,4.9> <0.70,256,3.5> <0.80,256,3.3> <0.70,256,3.5> <0.10,256,7.5> <1.20,256,5.0> <1.40,256,4.4> <1.40,-256,3.7> <0.50,256,3.8> <0.30,256,10.8> <0.40,256,21.3> <0.50,256,11.6> <0.50,256,4.0> <0.70,256,8.5> <1.20,256,9.1> <1.20,256,16.2> <1.00,256,5.0> <1.10,256,39.5> <1.40,228,28.2> <0.90,256,11.8> <0.70,256,7.3> <0.80,256,39.9> <1.00,256,17.9> <0.70,227,8.5> <1.10,227,56.0> s_ewma Keith Winstein and Hari Balakrishnan MIT Computer Science and Artificial Intelligence Laboratory TCP ex Machina: Computer-Generated Congestion Control

  60. Introduction The problem Remy Evaluation Discussion Split <0.90,256,10.2> <0.80,256,4.6> <1.00,-256,10.7> r_ewma <0.50,255,19.7> <1.40,254,28.5> <0.40,-256,55.4> <1.00,256,7.1> <1.40,252,74.6> <1.10,256,58.2> <1.80,256,3.1> <0.30,256,3.3> <1.70,254,3.3> <0.20,256,7.7> <1.10,256,4.2> <1.10,256,3.5> <1.60,256,7.4> <0.60,256,3.8> <0.80,256,3.0> <0.80,256,3.1> <0.10,256,4.8> <2.00,256,3.4> <1.80,256,6.6> <0.70,256,3.5> <0.90,256,3.2> <1.00,256,2.8> <0.30,256,6.0> <0.90,256,7.0> <1.50,256,5.3> <1.40,-256,4.6> <0.50,256,3.9> <0.40,256,10.0> <0.60,256,7.6> <0.40,256,20.5> <0.80,256,4.0> <0.50,256,6.1> <0.90,256,9.4> <0.60,256,16.0> <0.70,256,5.2> <1.00,256,39.3> <1.50,228,68.0> <0.90,256,10.3> <1.10,256,3.4> <1.00,256,39.0> <1.00,256,32.5> <0.70,227,14.8> <1.10,227,48.7> s_ewma Keith Winstein and Hari Balakrishnan MIT Computer Science and Artificial Intelligence Laboratory TCP ex Machina: Computer-Generated Congestion Control

  61. Introduction The problem Remy Evaluation Discussion Simulate <0.90,256,10.2> <0.80,256,4.6> <1.00,-256,10.7> r_ewma <0.50,255,19.7> <1.40,254,28.5> <0.40,-256,55.4> <1.00,256,7.1> <1.40,252,74.6> <1.10,256,58.2> <1.80,256,3.1> <0.30,256,3.3> <1.70,254,3.3> <0.20,256,7.7> <1.10,256,4.2> <1.10,256,3.5> <1.60,256,7.4> <0.60,256,3.8> <0.80,256,3.0> <0.80,256,3.1> <0.80,256,3.1> <0.10,256,4.8> <0.80,256,3.1> <0.80,256,3.1> <2.00,256,3.4> <1.80,256,6.6> <0.70,256,3.5> <0.90,256,3.2> <1.00,256,2.8> <0.30,256,6.0> <0.90,256,7.0> <1.50,256,5.3> <1.40,-256,4.6> <0.50,256,3.9> <0.40,256,10.0> <0.60,256,7.6> <0.40,256,20.5> <0.80,256,4.0> <0.50,256,6.1> <0.90,256,9.4> <0.60,256,16.0> <0.70,256,5.2> <1.00,256,39.3> <1.50,228,68.0> <0.90,256,10.3> <1.10,256,3.4> <1.00,256,39.0> <1.00,256,32.5> <0.70,227,14.8> <1.10,227,48.7> s_ewma Keith Winstein and Hari Balakrishnan MIT Computer Science and Artificial Intelligence Laboratory TCP ex Machina: Computer-Generated Congestion Control

  62. Introduction The problem Remy Evaluation Discussion Optimize <0.90,256,10.2> <0.80,256,4.6> <1.00,-256,10.7> r_ewma <0.50,255,19.7> <1.40,254,28.5> <0.40,-256,55.4> <1.00,256,7.1> <1.40,252,74.6> <1.10,256,58.2> <1.80,256,3.1> <0.30,256,3.3> <1.70,254,3.3> <0.20,256,7.7> <1.10,256,4.2> <1.10,256,3.5> <1.60,256,7.4> <0.60,256,3.8> <0.80,256,3.0> <0.80,256,3.1> <0.80,256,3.1> <0.10,256,4.8> <0.80,256,3.1> <0.80,256,3.1> <2.00,256,3.4> <1.80,256,6.6> <0.70,256,3.5> <0.90,256,3.2> <1.00,256,2.8> <0.30,256,6.0> <0.90,256,7.0> <1.50,256,5.3> <1.40,-256,4.6> <0.50,256,3.9> <0.40,256,10.0> <0.60,256,7.6> <0.40,256,20.5> <0.80,256,4.0> <0.50,256,6.1> <0.90,256,9.4> <0.60,256,16.0> <0.70,256,5.2> <1.00,256,39.3> <1.50,228,68.0> <0.90,256,10.3> <1.10,256,3.4> <1.00,256,39.0> <1.00,256,32.5> <0.70,227,14.8> <1.10,227,48.7> s_ewma Keith Winstein and Hari Balakrishnan MIT Computer Science and Artificial Intelligence Laboratory TCP ex Machina: Computer-Generated Congestion Control

  63. Introduction The problem Remy Evaluation Discussion Split <1.10,256,10.9> <0.80,256,3.5> <1.30,-256,5.0> r_ewma <0.50,255,19.7> <1.40,254,22.2> <0.40,-256,61.8> <1.00,256,6.9> <1.40,252,75.3> <1.10,256,57.4> <1.90,256,3.3> <0.30,256,3.1> <1.50,254,4.2> <0.20,256,8.2> <0.90,256,3.3> <0.90,256,3.2> <1.70,256,58.7> <0.20,256,3.2> <0.40,256,3.1> <0.70,256,3.3> <0.90,256,3.9> <0.30,256,4.1> <0.80,256,4.6> <0.80,256,3.2> <2.00,256,3.5> <2.00,256,7.2> <0.70,256,3.1> <0.90,256,3.1> <1.10,256,3.0> <0.20,256,5.0> <0.90,256,7.1> <1.40,256,4.7> <1.70,-256,5.5> <0.70,256,3.7> <0.40,256,10.9> <0.70,256,8.5> <0.20,256,14.8> <0.90,256,5.4> <0.80,256,5.1> <0.40,256,7.7> <0.80,256,9.5> <0.50,256,6.9> <0.70,256,39.8> <1.50,228,21.6> <0.90,256,12.6> <1.00,256,4.0> <1.00,256,38.1> <1.00,256,34.1> <0.70,227,8.5> <1.00,227,35.8> s_ewma Keith Winstein and Hari Balakrishnan MIT Computer Science and Artificial Intelligence Laboratory TCP ex Machina: Computer-Generated Congestion Control

  64. Introduction The problem Remy Evaluation Discussion Simulate <1.10,256,10.9> <0.80,256,3.5> <1.30,-256,5.0> r_ewma <0.50,255,19.7> <1.40,254,22.2> <0.40,-256,61.8> <1.00,256,6.9> <1.40,252,75.3> <1.10,256,57.4> <1.90,256,3.3> <0.30,256,3.1> <1.50,254,4.2> <0.20,256,8.2> <0.90,256,3.3> <0.90,256,3.2> <1.70,256,58.7> <0.20,256,3.2> <0.40,256,3.1> <0.70,256,3.3> <0.90,256,3.9> <0.30,256,4.1> <0.80,256,4.6> <0.80,256,3.2> <2.00,256,3.5> <2.00,256,7.2> <0.70,256,3.1> <0.90,256,3.1> <1.10,256,3.0> <0.20,256,5.0> <0.90,256,7.1> <1.40,256,4.7> <1.70,-256,5.5> <0.70,256,3.7> <0.40,256,10.9> <0.70,256,8.5> <0.90,256,5.4> <0.90,256,5.4> <0.20,256,14.8> <0.90,256,5.4> <0.90,256,5.4> <0.80,256,5.1> <0.40,256,7.7> <0.80,256,9.5> <0.50,256,6.9> <0.70,256,39.8> <1.50,228,21.6> <0.90,256,12.6> <1.00,256,4.0> <1.00,256,38.1> <1.00,256,34.1> <0.70,227,8.5> <1.00,227,35.8> s_ewma Keith Winstein and Hari Balakrishnan MIT Computer Science and Artificial Intelligence Laboratory TCP ex Machina: Computer-Generated Congestion Control

  65. Introduction The problem Remy Evaluation Discussion Optimize <1.10,256,10.9> <0.80,256,3.5> <1.30,-256,5.0> r_ewma <0.50,255,19.7> <1.40,254,22.2> <0.40,-256,61.8> <1.00,256,6.9> <1.40,252,75.3> <1.10,256,57.4> <1.90,256,3.3> <0.30,256,3.1> <1.50,254,4.2> <0.20,256,8.2> <0.90,256,3.3> <0.90,256,3.2> <1.70,256,58.7> <0.20,256,3.2> <0.40,256,3.1> <0.70,256,3.3> <0.90,256,3.9> <0.30,256,4.1> <0.80,256,4.6> <0.80,256,3.2> <2.00,256,3.5> <2.00,256,7.2> <0.70,256,3.1> <0.90,256,3.1> <1.10,256,3.0> <0.20,256,5.0> <0.90,256,7.1> <1.40,256,4.7> <1.70,-256,5.5> <0.70,256,3.7> <0.40,256,10.9> <0.70,256,8.5> <0.90,256,5.4> <0.90,256,5.4> <0.20,256,14.8> <0.90,256,5.4> <0.90,256,5.4> <0.80,256,5.1> <0.40,256,7.7> <0.80,256,9.5> <0.50,256,6.9> <0.70,256,39.8> <1.50,228,21.6> <0.90,256,12.6> <1.00,256,4.0> <1.00,256,38.1> <1.00,256,34.1> <0.70,227,8.5> <1.00,227,35.8> s_ewma Keith Winstein and Hari Balakrishnan MIT Computer Science and Artificial Intelligence Laboratory TCP ex Machina: Computer-Generated Congestion Control

  66. Introduction The problem Remy Evaluation Discussion Split <1.30,256,10.1> <0.80,256,3.7> <1.40,-256,50.7> r_ewma <0.50,255,19.7> <1.40,254,22.2> <0.40,-256,35.4> <0.80,256,4.6> <1.40,252,24.2> <1.10,256,57.4> <1.90,256,3.9> <0.30,256,4.0> <1.70,254,3.5> <0.20,256,14.7> <0.90,256,3.3> <1.10,256,3.1> <1.70,256,2.9> <0.30,256,4.8> <0.80,256,3.2> <0.80,256,3.1> <1.30,256,3.1> <0.20,256,3.9> <0.30,256,3.0> <0.70,256,3.3> <1.70,256,3.5> <2.00,256,8.7> <0.90,256,3.1> <0.40,256,3.3> <1.20,256,3.2> <0.20,256,5.2> <0.90,256,6.4> <1.10,256,4.7> <1.30,-256,5.4> <0.80,256,3.9> <0.70,256,10.1> <0.20,256,7.1> <0.80,256,6.1> <0.90,256,6.1> <0.40,256,14.8> <0.70,256,18.0> <0.60,256,5.1> <0.80,256,4.1> <0.70,256,7.0> <0.70,256,12.0> <0.30,256,7.7> <0.80,256,20.7> <1.70,228,10.3> <1.00,256,19.1> <1.10,256,11.1> <1.20,256,40.8> <1.00,256,22.2> <0.60,227,51.8> <1.00,227,29.9> s_ewma Keith Winstein and Hari Balakrishnan MIT Computer Science and Artificial Intelligence Laboratory TCP ex Machina: Computer-Generated Congestion Control

  67. Introduction The problem Remy Evaluation Discussion Simulate <1.30,256,10.1> <0.80,256,3.7> <1.40,-256,50.7> r_ewma <0.50,255,19.7> <1.40,254,22.2> <0.40,-256,35.4> <0.80,256,4.6> <1.40,252,24.2> <1.10,256,57.4> <1.90,256,3.9> <0.30,256,4.0> <1.70,254,3.5> <0.20,256,14.7> <0.90,256,3.3> <1.10,256,3.1> <1.70,256,2.9> <0.30,256,4.8> <0.80,256,3.2> <0.80,256,3.2> <0.80,256,3.2> <0.80,256,3.2> <0.80,256,3.1> <1.30,256,3.1> <0.20,256,3.9> <0.30,256,3.0> <0.70,256,3.3> <1.70,256,3.5> <2.00,256,8.7> <0.90,256,3.1> <0.40,256,3.3> <1.20,256,3.2> <0.20,256,5.2> <0.90,256,6.4> <1.10,256,4.7> <1.30,-256,5.4> <0.80,256,3.9> <0.70,256,10.1> <0.20,256,7.1> <0.80,256,6.1> <0.90,256,6.1> <0.40,256,14.8> <0.70,256,18.0> <0.60,256,5.1> <0.80,256,4.1> <0.70,256,7.0> <0.70,256,12.0> <0.30,256,7.7> <0.80,256,20.7> <1.70,228,10.3> <1.00,256,19.1> <1.10,256,11.1> <1.20,256,40.8> <1.00,256,22.2> <0.60,227,51.8> <1.00,227,29.9> s_ewma Keith Winstein and Hari Balakrishnan MIT Computer Science and Artificial Intelligence Laboratory TCP ex Machina: Computer-Generated Congestion Control

  68. Introduction The problem Remy Evaluation Discussion Optimize <1.30,256,10.1> <0.80,256,3.7> <1.40,-256,50.7> r_ewma <0.50,255,19.7> <1.40,254,22.2> <0.40,-256,35.4> <0.80,256,4.6> <1.40,252,24.2> <1.10,256,57.4> <1.90,256,3.9> <0.30,256,4.0> <1.70,254,3.5> <0.20,256,14.7> <0.90,256,3.3> <1.10,256,3.1> <1.70,256,2.9> <0.30,256,4.8> <0.80,256,3.2> <0.80,256,3.2> <0.80,256,3.2> <0.80,256,3.2> <0.80,256,3.1> <1.30,256,3.1> <0.20,256,3.9> <0.30,256,3.0> <0.70,256,3.3> <1.70,256,3.5> <2.00,256,8.7> <0.90,256,3.1> <0.40,256,3.3> <1.20,256,3.2> <0.20,256,5.2> <0.90,256,6.4> <1.10,256,4.7> <1.30,-256,5.4> <0.80,256,3.9> <0.70,256,10.1> <0.20,256,7.1> <0.80,256,6.1> <0.90,256,6.1> <0.40,256,14.8> <0.70,256,18.0> <0.60,256,5.1> <0.80,256,4.1> <0.70,256,7.0> <0.70,256,12.0> <0.30,256,7.7> <0.80,256,20.7> <1.70,228,10.3> <1.00,256,19.1> <1.10,256,11.1> <1.20,256,40.8> <1.00,256,22.2> <0.60,227,51.8> <1.00,227,29.9> s_ewma Keith Winstein and Hari Balakrishnan MIT Computer Science and Artificial Intelligence Laboratory TCP ex Machina: Computer-Generated Congestion Control

  69. Introduction The problem Remy Evaluation Discussion Split <1.20,256,8.5> <0.80,256,3.6> <1.30,-256,46.9> r_ewma <0.50,255,64.5> <1.40,254,14.2> <0.60,-256,35.6> <0.70,256,4.4> <1.40,252,24.2> <1.10,256,3.5> <1.80,256,3.2> <0.30,256,4.0> <1.70,254,3.5> <0.20,256,1.9> <0.90,256,4.1> <1.10,256,2.3> <1.70,256,10.9> <0.20,256,6.1> <0.60,256,3.5> <0.80,256,3.0> <0.80,256,3.8> <0.80,256,3.6> <0.90,256,3.3> <1.30,256,4.1> <0.20,256,3.2> <0.30,256,3.6> <0.80,256,3.5> <1.80,256,4.7> <1.90,256,6.5> <1.00,256,3.1> <0.50,256,3.8> <1.30,256,3.3> <0.20,256,5.3> <1.20,256,6.5> <1.10,256,5.4> <1.30,-256,6.3> <0.70,256,3.8> <0.80,256,9.7> <0.40,256,11.2> <0.60,256,5.8> <0.80,256,5.1> <0.30,256,13.9> <0.60,256,10.1> <0.70,256,5.4> <0.40,256,4.1> <0.70,256,6.7> <0.70,256,11.4> <0.60,256,8.8> <0.40,256,20.0> <1.70,228,15.7> <1.00,256,24.7> <0.90,256,17.6> <1.30,256,35.4> <0.90,256,9.2> <0.60,227,46.2> <1.00,227,37.1> s_ewma Keith Winstein and Hari Balakrishnan MIT Computer Science and Artificial Intelligence Laboratory TCP ex Machina: Computer-Generated Congestion Control

  70. Introduction The problem Remy Evaluation Discussion Simulate <1.20,256,8.5> <0.80,256,3.6> <1.30,-256,46.9> r_ewma <0.50,255,64.5> <1.40,254,14.2> <0.60,-256,35.6> <0.70,256,4.4> <1.40,252,24.2> <1.10,256,3.5> <1.80,256,3.2> <0.30,256,4.0> <1.70,254,3.5> <0.20,256,1.9> <0.90,256,4.1> <1.10,256,2.3> <1.70,256,10.9> <0.20,256,6.1> <0.60,256,3.5> <0.80,256,3.0> <0.80,256,3.8> <0.80,256,3.6> <0.90,256,3.3> <1.30,256,4.1> <0.20,256,3.2> <0.30,256,3.6> <0.80,256,3.5> <1.80,256,4.7> <1.90,256,6.5> <1.00,256,3.1> <0.50,256,3.8> <1.30,256,3.3> <0.20,256,5.3> <1.20,256,6.5> <1.10,256,5.4> <1.30,-256,6.3> <0.70,256,3.8> <0.70,256,3.8> <0.80,256,9.7> <0.70,256,3.8> <0.70,256,3.8> <0.40,256,11.2> <0.60,256,5.8> <0.80,256,5.1> <0.30,256,13.9> <0.60,256,10.1> <0.70,256,5.4> <0.40,256,4.1> <0.70,256,6.7> <0.70,256,11.4> <0.60,256,8.8> <0.40,256,20.0> <1.70,228,15.7> <1.00,256,24.7> <0.90,256,17.6> <1.30,256,35.4> <0.90,256,9.2> <0.60,227,46.2> <1.00,227,37.1> s_ewma Keith Winstein and Hari Balakrishnan MIT Computer Science and Artificial Intelligence Laboratory TCP ex Machina: Computer-Generated Congestion Control

  71. Introduction The problem Remy Evaluation Discussion Optimize <1.20,256,8.5> <0.80,256,3.6> <1.30,-256,46.9> r_ewma <0.50,255,64.5> <1.40,254,14.2> <0.60,-256,35.6> <0.70,256,4.4> <1.40,252,24.2> <1.10,256,3.5> <1.80,256,3.2> <0.30,256,4.0> <1.70,254,3.5> <0.20,256,1.9> <0.90,256,4.1> <1.10,256,2.3> <1.70,256,10.9> <0.20,256,6.1> <0.60,256,3.5> <0.80,256,3.0> <0.80,256,3.8> <0.80,256,3.6> <0.90,256,3.3> <1.30,256,4.1> <0.20,256,3.2> <0.30,256,3.6> <0.80,256,3.5> <1.80,256,4.7> <1.90,256,6.5> <1.00,256,3.1> <0.50,256,3.8> <1.30,256,3.3> <0.20,256,5.3> <1.20,256,6.5> <1.10,256,5.4> <1.30,-256,6.3> <0.70,256,3.8> <0.70,256,3.8> <0.80,256,9.7> <0.70,256,3.8> <0.70,256,3.8> <0.40,256,11.2> <0.60,256,5.8> <0.80,256,5.1> <0.30,256,13.9> <0.60,256,10.1> <0.70,256,5.4> <0.40,256,4.1> <0.70,256,6.7> <0.70,256,11.4> <0.60,256,8.8> <0.40,256,20.0> <1.70,228,15.7> <1.00,256,24.7> <0.90,256,17.6> <1.30,256,35.4> <0.90,256,9.2> <0.60,227,46.2> <1.00,227,37.1> s_ewma Keith Winstein and Hari Balakrishnan MIT Computer Science and Artificial Intelligence Laboratory TCP ex Machina: Computer-Generated Congestion Control

  72. Introduction The problem Remy Evaluation Discussion Split <1.30,256,8.5> <1.00,256,3.2> <1.20,-256,33.1> r_ewma <0.50,255,63.8> <1.40,254,8.6> <0.80,-256,23.6> <0.70,256,9.9> <1.50,252,18.5> <1.10,256,21.9> <1.80,256,3.1> <0.40,256,10.4> <1.70,254,3.5> <0.10,256,1.1> <1.00,256,3.7> <1.10,256,3.5> <1.70,256,5.2> <0.30,256,8.0> <0.60,256,3.4> <0.90,256,3.1> <0.80,256,4.0> <0.80,256,3.6> <0.80,256,3.4> <1.20,256,3.1> <0.20,256,5.1> <0.30,256,2.7> <0.80,256,3.3> <1.80,256,4.0> <1.90,256,4.8> <1.00,256,3.1> <0.60,256,3.9> <1.20,256,5.7> <0.40,256,4.1> <1.10,256,6.2> <1.40,256,3.6> <1.60,-256,5.7> <1.20,256,3.5> <0.70,256,4.0> <1.10,256,11.5> <0.80,256,4.6> <0.80,256,4.6> <0.30,256,12.0> <0.60,256,4.9> <0.80,256,11.4> <0.20,256,13.6> <0.70,256,11.7> <0.70,256,5.4> <0.40,256,5.5> <0.50,256,7.3> <0.60,256,10.5> <0.50,256,7.1> <0.50,256,20.7> <1.70,228,16.5> <0.80,256,24.8> <0.70,256,11.2> <1.30,256,41.9> <0.90,256,10.0> <0.60,227,60.8> <1.00,227,57.1> s_ewma Keith Winstein and Hari Balakrishnan MIT Computer Science and Artificial Intelligence Laboratory TCP ex Machina: Computer-Generated Congestion Control

  73. Introduction The problem Remy Evaluation Discussion Simulate <1.30,256,8.5> <1.00,256,3.2> <1.20,-256,33.1> r_ewma <0.50,255,63.8> <1.40,254,8.6> <0.80,-256,23.6> <0.70,256,9.9> <1.50,252,18.5> <1.10,256,21.9> <1.80,256,3.1> <0.40,256,10.4> <1.70,254,3.5> <0.10,256,1.1> <1.00,256,3.7> <1.10,256,3.5> <1.70,256,5.2> <0.30,256,8.0> <0.60,256,3.4> <0.90,256,3.1> <0.80,256,4.0> <0.80,256,3.6> <0.80,256,3.4> <1.20,256,3.1> <0.20,256,5.1> <0.30,256,2.7> <0.80,256,3.3> <1.80,256,4.0> <1.90,256,4.8> <1.00,256,3.1> <0.60,256,3.9> <1.20,256,5.7> <0.40,256,4.1> <1.10,256,6.2> <1.40,256,3.6> <1.60,-256,5.7> <1.20,256,3.5> <0.70,256,4.0> <1.10,256,11.5> <0.80,256,4.6> <0.80,256,4.6> <0.30,256,12.0> <0.60,256,4.9> <0.80,256,11.4> <0.20,256,13.6> <0.70,256,11.7> <0.70,256,5.4> <0.50,256,7.3> <0.50,256,7.3> <0.40,256,5.5> <0.60,256,10.5> <0.50,256,7.3> <0.50,256,7.3> <0.50,256,7.1> <0.50,256,20.7> <1.70,228,16.5> <0.80,256,24.8> <0.70,256,11.2> <1.30,256,41.9> <0.90,256,10.0> <0.60,227,60.8> <1.00,227,57.1> s_ewma Keith Winstein and Hari Balakrishnan MIT Computer Science and Artificial Intelligence Laboratory TCP ex Machina: Computer-Generated Congestion Control

  74. Introduction The problem Remy Evaluation Discussion Optimize <1.30,256,8.5> <1.00,256,3.2> <1.20,-256,33.1> r_ewma <0.50,255,63.8> <1.40,254,8.6> <0.80,-256,23.6> <0.70,256,9.9> <1.50,252,18.5> <1.10,256,21.9> <1.80,256,3.1> <0.40,256,10.4> <1.70,254,3.5> <0.10,256,1.1> <1.00,256,3.7> <1.10,256,3.5> <1.70,256,5.2> <0.30,256,8.0> <0.60,256,3.4> <0.90,256,3.1> <0.80,256,4.0> <0.80,256,3.6> <0.80,256,3.4> <1.20,256,3.1> <0.20,256,5.1> <0.30,256,2.7> <0.80,256,3.3> <1.80,256,4.0> <1.90,256,4.8> <1.00,256,3.1> <0.60,256,3.9> <1.20,256,5.7> <0.40,256,4.1> <1.10,256,6.2> <1.40,256,3.6> <1.60,-256,5.7> <1.20,256,3.5> <0.70,256,4.0> <1.10,256,11.5> <0.80,256,4.6> <0.80,256,4.6> <0.30,256,12.0> <0.60,256,4.9> <0.80,256,11.4> <0.20,256,13.6> <0.70,256,11.7> <0.70,256,5.4> <0.50,256,7.3> <0.50,256,7.3> <0.40,256,5.5> <0.60,256,10.5> <0.50,256,7.3> <0.50,256,7.3> <0.50,256,7.1> <0.50,256,20.7> <1.70,228,16.5> <0.80,256,24.8> <0.70,256,11.2> <1.30,256,41.9> <0.90,256,10.0> <0.60,227,60.8> <1.00,227,57.1> s_ewma Keith Winstein and Hari Balakrishnan MIT Computer Science and Artificial Intelligence Laboratory TCP ex Machina: Computer-Generated Congestion Control

  75. Introduction The problem Remy Evaluation Discussion Split <1.40,256,8.3> <1.00,256,3.0> <1.40,-256,27.5> r_ewma <0.50,255,69.4> <1.40,254,14.3> <0.70,-256,29.0> <0.90,256,2.8> <1.50,252,18.5> <1.10,256,73.9> <1.90,256,3.1> <0.50,256,3.3> <1.70,254,3.2> <0.20,256,58.7> <1.00,256,3.8> <1.00,256,5.9> <1.80,256,2.8> <0.30,256,9.8> <0.50,256,3.4> <1.00,256,2.8> <0.90,256,3.4> <0.80,256,2.9> <0.60,256,3.3> <1.20,256,3.1> <0.20,256,3.5> <0.50,256,4.4> <0.80,256,4.0> <1.50,256,3.1> <1.90,256,5.6> <1.10,256,3.2> <0.90,256,3.6> <1.40,256,5.0> <0.30,256,3.8> <1.00,256,5.6> <1.30,256,3.5> <2.00,-256,4.2> <1.20,256,3.9> <0.80,256,5.1> <1.30,256,9.9> <0.50,256,12.1> <0.80,256,4.9> <0.40,256,18.3> <0.80,256,4.1> <1.30,256,17.9> <0.20,256,13.7> <1.00,256,11.3> <0.70,256,4.6> <0.40,256,6.6> <0.90,256,8.7> <0.40,256,4.9> <0.50,256,11.1> <0.50,256,7.2> <0.80,256,8.0> <0.50,256,8.6> <0.30,256,21.5> <1.70,228,61.2> <0.80,256,19.1> <0.80,256,9.8> <1.20,256,40.9> <0.90,256,10.3> <0.60,227,21.6> <1.00,227,57.1> s_ewma Keith Winstein and Hari Balakrishnan MIT Computer Science and Artificial Intelligence Laboratory TCP ex Machina: Computer-Generated Congestion Control

Recommend

Internet congestion control: TCP Internet congestion control: TCP 1988: &quot;Congestion

Internet congestion control: TCP Internet congestion control: TCP 1988: &quot;Congestion

Internet congestion control: TCP Internet congestion control: TCP 1988: &quot;Congestion Avoidance and Control&quot; (Jacobson/Karels) Combined congestion/flow control for TCP Goal: stability - in equilibrum, no packet is sent into the

482 views • 24 slides
1 TCP AIMD TCP Congestion Control additive increase: multiplicative decrease: increase CongWin

1 TCP AIMD TCP Congestion Control additive increase: multiplicative decrease: increase CongWin

Principles of Congestion Control Congestion: informally: too many sources sending too much Congestion Control data too fast for network to handle different from flow control! manifestations: lost packets (buffer overflow at

242 views • 3 slides
Advanced Congestion Control (Hosts) 14-740: Fundamentals of Computer Networks Bill Nace

Advanced Congestion Control (Hosts) 14-740: Fundamentals of Computer Networks Bill Nace

Advanced Congestion Control (Hosts) 14-740: Fundamentals of Computer Networks Bill Nace Material from Computer Networking: A Top Down Approach, J.F. Kurose and K.W. Ross Congestion Control (2) Apply some control theory Region 1: Low

546 views • 32 slides
Congestion Control Outline Queuing Discipline Reacting to Congestion Avoiding Congestion 1

Congestion Control Outline Queuing Discipline Reacting to Congestion Avoiding Congestion 1

Congestion Control Outline Queuing Discipline Reacting to Congestion Avoiding Congestion 1 Issues Two sides of the same coin pre-allocate resources to avoid congestion (e.g. telephone networks) control congestion if (and when)

494 views • 14 slides
What do you mean, Congestion? some history Congestion Collapse

What do you mean, Congestion? some history Congestion Collapse

What do you mean, Congestion? some history Congestion Collapse Congestion Avoidance Congestion Control Explicit Congestion Notification Datagram Congestion Control Protocol this

426 views • 15 slides
End-to-End Previously on CS5229 , Not everyone uses TCP Congestion Control TCP Congestion

End-to-End Previously on CS5229 , Not everyone uses TCP Congestion Control TCP Congestion

End-to-End Previously on CS5229 , Not everyone uses TCP Congestion Control TCP Congestion Control UDP: Why not congestion controlled? Media streaming 1. UDP has low delay, no Gaming need full reliability VoIP

280 views • 12 slides
CompSci 514: Computer Networks Lecture 5: Congestion Control Xiaowei Yang 1 Outline

CompSci 514: Computer Networks Lecture 5: Congestion Control Xiaowei Yang 1 Outline

CompSci 514: Computer Networks Lecture 5: Congestion Control Xiaowei Yang 1 Outline Background on TCP congestion control The congestion control algorithm Theoretic background Macroscopic modeling 2 The Internet

662 views • 38 slides
TCP TCP Congestion Control Congestion Control Essential strategy :: The TCP host sends

TCP TCP Congestion Control Congestion Control Essential strategy :: The TCP host sends

11/19/2009 TCP Congestion Control TCP Congestion Control TCP TCP Congestion Control Congestion Control Essential strategy :: The TCP host sends packets into the network without a reservation and then the host reacts to observable

823 views • 6 slides
The Present and Future of Congestion Control Mark Handley Outline Purpose of congestion

The Present and Future of Congestion Control Mark Handley Outline Purpose of congestion

The Present and Future of Congestion Control Mark Handley Outline Purpose of congestion control The Present: TCPs congestion control algorithm (AIMD) TCP-friendly congestion control for multimedia Datagram Congestion

984 views • 54 slides
CS 356: Computer Network Architectures Lecture 20: Congestion Avoidance Chap. 6.4 and related

CS 356: Computer Network Architectures Lecture 20: Congestion Avoidance Chap. 6.4 and related

CS 356: Computer Network Architectures Lecture 20: Congestion Avoidance Chap. 6.4 and related papers Xiaowei Yang xwy@cs.duke.edu TCP Congestion Control History The original TCP/IP design did not include congestion control and avoidance

685 views • 33 slides
TCP/IP Over Lossy Links - TCP SACK without Congestion Control Organization 1. The History of

TCP/IP Over Lossy Links - TCP SACK without Congestion Control Organization 1. The History of

TCP/IP Over Lossy Links - TCP SACK without Congestion Control Organization 1. The History of TCP 2.Current TCP Congestion Control 3.Design Ideas: no congestion control at all 4.Measurement Results 5.Future TCP Congestion Control?

464 views • 10 slides
Sliding Window Protocol and TCP Congestion Control Simon S. Lam Department of Computer Science

Sliding Window Protocol and TCP Congestion Control Simon S. Lam Department of Computer Science

Sliding Window Protocol and TCP Congestion Control Simon S. Lam Department of Computer Science Th The University of Texas at Austin U i it f T t A ti TCP Congestion Control (Simon S. Lam) 1 1 Reliable data transfer f important in

958 views • 41 slides
Convex Optimization and Congestion Control - Part I (aka The Kelly Approach to Congestion

Convex Optimization and Congestion Control - Part I (aka The Kelly Approach to Congestion

Convex Optimization and Congestion Control - Part I (aka The Kelly Approach to Congestion Control) Laila Daniel and Krishnan Narayanan 11th March 2013 Motivation The Kelly approach exposes THE PROBLEM underlying any congestion control

250 views • 24 slides
Congestion Control Mark Handley Outline Part 1: Traditional congestion control for bulk

Congestion Control Mark Handley Outline Part 1: Traditional congestion control for bulk

Congestion Control Mark Handley Outline Part 1: Traditional congestion control for bulk transfer. Part 2: Congestion control for realtime traffic. Part 3: High-speed congestion control. Part 1 Traditional congestion control for

1.23k views • 98 slides
CS 514: Computer Networks Lecture 6: Explicit Congestion Control Xiaowei Yang Review A

CS 514: Computer Networks Lecture 6: Explicit Congestion Control Xiaowei Yang Review A

CS 514: Computer Networks Lecture 6: Explicit Congestion Control Xiaowei Yang Review A fundamental question to networking: Multiple users share the same network. Who can send at what speed? Approach 1: End-to-end congestion

340 views • 30 slides
Machina Research Where is the value in IoT? IoT data and analytics may have an answer Emil

Machina Research Where is the value in IoT? IoT data and analytics may have an answer Emil

Machina Research Where is the value in IoT? IoT data and analytics may have an answer Emil Berthelsen, Principal Analyst April 28, 2016 About Machina Research Machina Research is the worlds leading provider of market intelligence and

521 views • 16 slides
Internet Congestion Control Research Group Mark Handley UCL Congestion Control The Internet

Internet Congestion Control Research Group Mark Handley UCL Congestion Control The Internet

Internet Congestion Control Research Group Mark Handley UCL Congestion Control The Internet only functions because TCPs congestion control does an effective job of matching traffic demand to available capacity. TCPs Window Time

628 views • 14 slides
1 ACK clocking ACK clocking ACK clocking spreads out bursts ACK clocking spreads out bursts

1 ACK clocking ACK clocking ACK clocking spreads out bursts ACK clocking spreads out bursts

TCP Congestion Control Review Congestion Control (contd) Congestion control consists of 3 tasks Detect congestion Adjust sending rate Determine available bandwidth How does TCP do each of these? Packets vs. Bytes TCP Start

491 views • 8 slides
Lecture 8 Congestion Control EECS 122 University of California Berkeley TOC: Congestion

Lecture 8 Congestion Control EECS 122 University of California Berkeley TOC: Congestion

Lecture 8 Congestion Control EECS 122 University of California Berkeley TOC: Congestion Control The Problem Questions Approaches TCP: Algorithm TCP Refinements Summary EECS 122 Walrand 2 Congestion Control: The Problem Flows share

487 views • 30 slides
CS 514: Computer Networks Lecture 7: Other Congestion Control Algorithms Xiaowei Yang Overview

CS 514: Computer Networks Lecture 7: Other Congestion Control Algorithms Xiaowei Yang Overview

CS 514: Computer Networks Lecture 7: Other Congestion Control Algorithms Xiaowei Yang Overview Other congestion algorithms TCP weaknesses Slow convergence for large bandwidth-delay product networks Unfairness among flows

479 views • 34 slides
Layer Optimization: Congestion Control CS 118 Computer Network Fundamentals Peter Reiher

Layer Optimization: Congestion Control CS 118 Computer Network Fundamentals Peter Reiher

Layer Optimization: Congestion Control CS 118 Computer Network Fundamentals Peter Reiher Lecture 17 CS 118 Page 1 Winter 2016 We can lose packets for many reasons Corruption Not delivered to receiver Poor flow control But

1.07k views • 65 slides
Convex Optimization Congestion Control Laila Daniel and Krishnan Narayanan 11th March 2013

Convex Optimization Congestion Control Laila Daniel and Krishnan Narayanan 11th March 2013

Convex Optimization Congestion Control Laila Daniel and Krishnan Narayanan 11th March 2013 Congestion Control Two fundamental papers in Internet Congestion Control V. Jacobson. Congestion avoidance and control. Proceeding SIGCOMM

278 views • 16 slides
On the Limitation of Fluid-based Approach for Internet Congestion Control Do Young Eun

On the Limitation of Fluid-based Approach for Internet Congestion Control Do Young Eun

On the Limitation of Fluid-based Approach for Internet Congestion Control Do Young Eun http://www4.ncsu.edu/~dyeun Dept. of Electrical and Computer Engineering North Carolina State University Oct. 18, 2005 Outline TCP/AQM Congestion

840 views • 27 slides
AC DC TCP: Virtual Congestion Control Enforcement for Datacenter Networks Ke Keqiang He He ,

AC DC TCP: Virtual Congestion Control Enforcement for Datacenter Networks Ke Keqiang He He ,

AC DC TCP: Virtual Congestion Control Enforcement for Datacenter Networks Ke Keqiang He He , Eric Rozner, Kanak Agarwal, Yu Gu, Wes Felter, John Carter, Aditya Akella 1 Datacenter Network Congestion Control Congestion is not rare in

560 views • 34 slides

More recommend