MPTCP is not Pareto-Optimal

1. MPTCP is not Pareto-Optimal Performance Issues and a possible solution B98505024吳昇峰

2. 帕累托效率（Pareto Efficiency） • 帕累托最優是指資源分配的一種理想狀態 • 固有的一群人和可分配的資源，從一種分配狀態到另一種狀態的變化中，在沒有使任何人境況變壞的前提下，使得至少一個人變得更好， • 帕累托最優是公平與效率的“理想王國”。

3. TCP • Use a window base congestion control • Disadvantages • Increase a user output↑ must decrease another user↓ or increase congestion cost↑ • Try to build Multipath transport protocol • Use the best paths available to users Disadvantages • Fail quickly detect free capacity • Exhibit flappiness→multiple good paths →randomly flip its traffic between them.

4. User expectations

5. What technology provides 3G celltower IP 1.2.3.4

6. What technology provides 3G celltower IP 1.2.3.4 IP 5.6.7.8

7. What technology provides 3G celltower IP 1.2.3.4 IP 5.6.7.8 When IP addresses change TCP connectionshave to be re-established !

8. MPTCP: Multi-path Transport Solution of IETF • allows a user to split its traffic across multiple paths; improve reliability and throughput • challenge: design of congestion control algorithm

9. "Linked-Increases Algorithm" • Follow on adhoc design • adhocdesign based on 3 goals • improve throughput:total throughput ≥ TCP over best path • do not harm: not more aggressive than a TCP over a path • balance congestionwhile meeting the first two goals • as also stated in RFC 6356, LIA does not fully satisfy goal 3

10. 根據pr-1/ϵpr is the loss probability (at 傳輸率r) • ϵ=2 used uncoupled TCP • ϵ=0 sent only the best path (flappy) • At most increase 1/wr

11. LIA’s design forces tradeoff between responsiveness and load balancing provide load balancing be responsive responsive but bad load balancing optimal load balacing but not responsive LIA’s implementation (RFC 6356) εis a design parameter ε=0 ε=2 ε=1

12. We identified problems with the currentMPTCP implementation • P1: MPTCP can penalize users • Upgrade some TCP to MPTCP can reduce throughput of other users without any improvement • P2: MPTCP users are excessively aggressive toward TCP users • P1 and P2 are attributed to LIA (linked increase) • LIAfails to every design goals especially satisfy goal 3

13. Measurement-based study supported by theory

14. OLIA • It is Pareto-optimal • Solve problem p1 and p2 • It is not flappy • Has better responsiveness

15. MPTCP CAN PENALIZE USERS upgrading some TCP users to MPTCP can reduce the throughput of others without any benefit to the upgraded users

16. Scenario A: MPTCP can penalize TCP users bottleneck for Type 1users is at server side high speed connections N1 C1 N1 x1 bottleneck for Type 2 users is at access side

17. Scenario A: MPTCP can penalize TCP users bottleneck for Type 1users is at server side high speed connections N1 C1 x2 N1(x1+x2 ) bottleneck for Type 2 users is at access side

18. Throughput of type 2 users reduced without any benefit for type 1 users x2 N1 C1

19. We compare MPTCP with a theoretical baselines • optimal algorithm with probing cost:theoretical optimal load balancing including minimal probing traffic • using a windows-based algorithm, a minimum probing traffic of 1 MSS/RTT is sent over each path

20. Part of problem is in nature of things, but MPTCP seems to be far from optimal x2 N1 C1

21. Scenario B: MPTCP can penalize other MPTCP users bottleneck with capacity Cx bottleneck with capacity CT

22. By upgrading red users to MPTCP, the throughput of everybody decreases decrease is 3% using optimal algorithm with probing cost Use LIA when CX/CT ≈0.75, the Blue users decrease by up to 20%.

23. Scenario C: MPTCP users could be excessively aggressive towards regular TCP users (P2) wifi

24. Scenario C: OLIA achieves much better fairness (solving P2) when C1/C2≥1, for any theoretical fairness criterion: (x1+x2)/C1=1and y/C2=1

25. CAN THE SUBOPTIMALITY OF MPTCP WITH LIA BE FIXED?

26. OLIA: an algorithm inspired by utility maximization framework • simultaneously provides responsiveness and load balancing • an adjustment of optimal algorithm • by adapting windows increases as a function of quality of paths, we make it responsive and non-flappy • implemented on the MPTCP Linux kernel

27. Definition: set of best paths and paths with maximum windows • for a user u, with Ru as set of paths, define • set of best paths: • set of path with maximum windows: • : number of successful transmissions between losses on path r • rttr(t): RTT on path r

28. OLIA: "Opportunistic Linked-Increases Algorithm" For a user u, on each path r in Ru: • increase part: for each ACK on r, increase wr by αr= • decrease part: each loss on r, decreases wr by wr/2 optimal load balancing responsiveness OLIA increases windows faster on the paths that are the best but have small windows. Increase is slower on the paths with maximum windows

29. An illustrative example of OLIA’s behavior MPTCP with OLIA MPTCP with OLIA MPTCP with LIA MPTCP with LIA OLIAuse both of paths.and no sign of flappiness second path is congested, OLIA uses only the first one

30. Theoretical results: OLIA solves problems P1 and P2 • using a fluid model of OLIA • Theorem: OLIA satisfies design goals of LIA • Theorem: OLIA is Pareto optimal • Theorem: when all paths of a user have similar RTTs, OLIA provides optimal load balancing

31. Scenario A: OLIA performs close to optimal algorithm with probing cost x2 N1 C1

32. Loss probability

33. Scenario B: using OLIA, we observe 3.5% drop in aggregate throughput MPTCP with OLIA MPTCP with LIA 15 blue and 15 red users, CX=27 and CT=36 Mbps

34. Scenario C:

35. Summary • MPTCP with LIA suffers from important performance problems • these problems can be mitigated in practice • OLIA: inspired by utility maximization framework

36. Question?