Coupled conges-on control for RTP media dra$-­‐welzl-­‐rmcat-­‐coupled-­‐cc-­‐04 Michael Welzl, Safiqul Islam, Stein Gjessing RMCAT 91st IETF Mee/ng Honolulu, US 12 November 2014 1 Context, background •  Having mul/ple conges/on controlled flows from the same sender compete on the same boLleneck is detrimental –  By first combining their conges/on controllers in the sender, we can beLer control fairness (with priori/es) and get less delay and loss •  Two elements: 1) shared boLleneck detec/on (sbd), 2) coupled conges/on control –  In rtcweb, 1) can some/mes be very easy: same 6-­‐
tuple. But measurement-­‐based sbd enables broader applica/on of 2) (same sender, different receivers) 2 What’s new in -­‐04? •  Minor update –  Extended SBD sec/on –  Added a note about window based conges/on controllers •  Papers, code etc. available via: hLp://heim.ifi.uio.no/safiquli/coupled-­‐cc/index.html 3 Shared boLlenecks •  Coupled CC only makes sense across a common boLleneck –  This was ignored in prior work –  But how to know? 1.  Mul/plexing (same 5-­‐, actually 6-­‐tuple) –  but only for same source/des/na/on hosts 2.  Configura/on (e.g. common wireless uplink) 3.  Measurement –  Never 100% reliable, but: different receivers possible! –  Historically considered imprac/cal, but recent work: David Hayes, Simone Ferlin-­‐Oliveira, Michael Welzl: "PracJcal Passive Shared BoLleneck DetecJon Using Shape Summary StaJsJcs", IEEE LCN 2014, 8-­‐11 September 2014
4 Some simula/on results F1 F1 F2 F2 Fn •
•
•
•
R1 R2 Fn BoLleneck – 10 Mbps Packet Size – 1500 Bytes RTT – 100 ms All tests carried out 10 /mes with random start /mes picked from first second 5 85
100
80
90
Link Utilization %
Average Queue Length
Mul/ple LEDBAT flows 75
70
65
60
FSE
Without FSE
2
4
6
8
10
12
14
# of Flows
Average Queue Length of LEDBAT Flows – with and without FSE 16
18
20
80
70
60
50
FSE
Without FSE
2
4
6
8
10
12
14
16
18
20
# of Flows
Link U/liza/on of LEDBAT Flows – with and without FSE 6 Next step •  Working with SCTP and LEDBAT •  Working with one of the RMCAT conges/on controllers 7 Q&A 8