MPTCP is not Pareto-Optimal

Performance Issues and a possible solution

B98505024吳昇峰

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”。

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 usersDisadvantages• Fail quickly detect free capacity• Exhibit flappiness →multiple good paths →randomly flip its traffic between

them.

User expectations

What technology provides3G celltower

IP 1.2.3.4

What technology provides3G celltower

IP 1.2.3.4

IP 5.6.7.8

What technology provides3G celltower

When IP addresses change TCP connectionshave to be re-established !

IP 1.2.3.4

IP 5.6.7.8

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 algorithm8

"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

•根據 pr-1/ϵ pr 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

LIA’s design forces tradeoff between responsiveness and load balancing

provide load balancing be 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

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)• LIA fails to every design goals especially satisfy goal 3

12

Measurement-based study supported by theory

13

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

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

15

Scenario A: MPTCP can penalize TCP users

16bottleneck for Type 2

users is at access side

bottleneck for Type 1users is at server side

high speed connections

N1 C1

N1 x1

Scenario A: MPTCP can penalize TCP users

17bottleneck for Type 2

users is at access side

bottleneck for Type 1users is at server side

high speed connections

N1 C1

N1( x1+x2 )x2

Throughput of type 2 users reduced without any benefit for type 1 users

18

N1 C1x2

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

Part of problem is in nature of things, but MPTCP seems to be far from optimal

20

N1 C1x2

Scenario B: MPTCP can penalize other MPTCP users

21

bottleneck with capacity Cx

bottleneck with capacity CT

By upgrading red users to MPTCP, the throughput of everybody decreases

22

decrease is 3% using optimal algorithm with probing cost Use LIA when CX /CT ≈0.75, the Blue users decrease by

up to 20%.

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

23

wifi

Scenario C: OLIA achieves much better fairness (solving P2)

24

when C1/C2≥1, for any theoretical fairness criterion: (x1+x2)/C1=1 and y/C2=1

CAN THE SUBOPTIMALITY OF MPTCP WITH LIA BE FIXED?

25

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

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

27

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/228

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

An illustrative example of OLIA’s behavior

29OLIA use both of paths.and no sign

of flappinesssecond path is congested,

OLIA uses only the first one

MPTCP with LIA

MPTCP with OLIA

MPTCP with LIA

MPTCP with OLIA

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

Scenario A: OLIA performs close to optimal algorithm with probing cost

31

N1 C1x2

Loss probability

Scenario B: using OLIA, we observe 3.5% drop in aggregate throughput

3315 blue and 15 red users, CX=27 and CT=36 Mbps

MPTCP with OLIA

MPTCP with LIA

Scenario C:

34

Summary•MPTCP with LIA suffers from important

performance problems

• these problems can be mitigated in practice• OLIA: inspired by utility maximization framework

35

Question?