Presentation is loading. Please wait.

Presentation is loading. Please wait.

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

Similar presentations


Presentation on theme: "MPTCP is not Pareto- Optimal Performance Issues and a possible solution B98505024 吳昇峰."— Presentation transcript:

1 MPTCP is not Pareto- Optimal Performance Issues and a possible solution B 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

6 What technology provides 3G celltower IP IP

7 What technology provides 3G celltower When IP addresses change TCP connections have to be re-established ! IP IP

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 8

9 "Linked-Increases Algorithm" Follow on adhoc design adhoc design based on 3 goals 1.improve throughput: total throughput ≥ TCP over best path 2.do not harm: not more aggressive than a TCP over a path 3.balance congestion while meeting the first two goals as also stated in RFC 6356, LIA does not fully satisfy goal 3 9

10 根據 p r -1/ϵ p r 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/w r

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

12 We identified problems with the current MPTCP 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) LIA fails to every design goals especially satisfy goal 3 12

13 Measurement-based study supported by theory 13

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 15

16 Scenario A: MPTCP can penalize TCP users 16 bottleneck for Type 2 users is at access side bottleneck for Type 1 users is at server side high speed connections N 1 C 1 N 1 x 1

17 Scenario A: MPTCP can penalize TCP users 17 bottleneck for Type 2 users is at access side bottleneck for Type 1 users is at server side high speed connections N 1 C 1 N 1 ( x 1 +x 2 ) x2x2

18 Throughput of type 2 users reduced without any benefit for type 1 users 18 N 1 C 1 x2x2

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 19

20 Part of problem is in nature of things, but MPTCP seems to be far from optimal 20 N 1 C 1 x2x2

21 Scenario B: MPTCP can penalize other MPTCP users 21 bottleneck with capacity C x bottleneck with capacity C T

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

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

24 Scenario C: OLIA achieves much better fairness (solving P2) 24 when C 1 /C 2 ≥1, for any theoretical fairness criterion: (x1+x2)/C 1 =1 and y/C 2 =1

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

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 26

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

28 OLIA: "Opportunistic Linked-Increases Algorithm" For a user u, on each path r in R u : increase part: for each ACK on r, increase w r by α r = decrease part: each loss on r, decreases w r by w r /2 28 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 29 OLIA use both of paths.and no sign of flappiness second path is congested, OLIA uses only the first one MPTCP with LIA MPTCP with OLIA MPTCP with LIA MPTCP with OLIA

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 30

31 Scenario A: OLIA performs close to optimal algorithm with probing cost 31 N 1 C 1 x2x2

32 Loss probability

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

34 Scenario C: 34

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

36 Question?


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

Similar presentations


Ads by Google