Download presentation

Presentation is loading. Please wait.

Published byJaden McNamara Modified over 4 years ago

1
ARC TCP Workshop, ENS, Paris, November 5-7, 2003 Equation-Based Rate Control: Is it TCP-friendly ? Milan Vojnovic Joint work with Jean-Yves Le Boudec

2
2 The Axiom: TCP-friendliness Requires adaptive sources to obey to TCP in the following sense: TCP-friendliness (late 1990s) A flow that is not TCP-friendly is one whose long-term arrival rate exceeds that of any conformant TCP in the same circumstances. Floyd and Fall, 1999

3
3 Equation-Based Rate Control: Basic Control Estimator of 1/p: Send rate: Example Protocol: TFRC (RFC 3448, IETF proposed standard, Jan 2003)

4
4 Is Equation-Based Rate Control a TCP Friend ? We deduce: the Engineering Intuition Problem: When the Intuition is True and when Not ? p -> f(p) is TCP loss-throughput formula So, it must be that if I adjust the send rate at loss-events to f(), evaluated at the on-line estimated loss-event rate, my new protocol will be TCP-friendly

5
5 Outline 1. Breakdown the TCP-friendliness into sub-conditions, study the sub-conditions separately oWhy the common evaluation practice to verify TCP- friendliness is not good ? 2. TCP-friendliness is difficult to verify oCounterexamples to TCP-friendliness 3. Conservativeness is easier oSufficient conditions for conservativeness oOr bounded non-conservativeness

6
6 1. Common Evaluation Practice Non-TCP Common Practice: TCP Why the common evaluation practice is NOT GOOD ? - hides a cause of the observed throughput deviation - may lead a protocol designer to an improper adjustment measured throughputs x x Test: TCP-friendly iff x <= x

7
7 Breakdown the TCP-Friendliness Condition (I) Does the source verify x <= f(p,r) ? (II) Does the source attain the same loss-event rate as TCP ? (III)Does the source see the same average round-trip time as TCP ? (IV)Does TCP verify its throughput formula ? Important to BREAKDOWN the TCP-friendliness condition into sub-conditions, and study them separately !

8
8 Breakdown the TCP-Friendliness Condition (Contd) (I) Conservativeness x <= f(p, r) (II) Loss-Event Rates p >= p (III) Round-Trip Times r >= r (IV) Obedience of TCP to the Formula x >= f(p, r) If (I), (II), (III), and (IV) hold, that implies TCP-friendliness. TCP Equation-Based Rate Control (x, p, r) throughputloss-event rateaverage RTT

9
9 2. Counterexample to TCP-Friendliness: AIMD experiences larger loss rate than EBRC EBRC r Ass. EBRC uses f(p) in (1) AIMD (a,b) r (1) TCP-like (b=1/2) p/p=16/9 (approx. 1.7778) Ob: p > p non-TCP-friendliness Example 1: Either One AIMD or One EBRC over a Link

10
10 Convergence for One EBRC over a Link slope K 2 /2

11
11 Convergence for One EBRC over a Link (Contd) Can be seen as Jacobi iterative solving of: The equilibrium point: If stable: Remarks oboth AIMD and EBRC are rate-based oboth AIMD and EBRC are fluid, no packetization effects => the deviation of the loss-event rates is intrinsic to the very nature of the dynamics of the two controls

12
12 Validation by ns-2 Simulation b pakets TFRC b pakets TCP x/x x/f(p,r)p/pr/rx/f(p,r) b Breakdown:

13
13 AIMD sees larger loss rate than EBRC (Contd) otime t is a loss-event iff at t- the sum of the send rates of the two sources = r oa loss-event is assigned to either AIMD or EBRC oZ n = 1 iff the nth loss-event is assigned to EBRC, else Z n =0 g : R + L+1 -> R + is a non-linear function; the system is non-linear Example 2: One AIMD and One EBRC Competing for a Link

14
14 Example 2: Numerical Simulations

15
15 Example 2: Validation by ns-2 Simulation b pakets TCP TFRC x/x b x/f(p,r)p/pr/rx/f(p,r) Breakdown:

16
16 Internet Measurements INRIA, KTH, UMASS,UMELB EPFL Long-lived transmissions with TFRC and TCP Estimated: loss-event rates, average round-trip times, throughputs

17
17 EPFL to UMASS x/x TFRC/TCP throughput x/f(p,r)p/pr/rx/f(p,r) Breakdown into Sub-Conditions:

18
18 3. Conservativeness Convergence: oThe send rate control: oThe estimator is updated at special points in time Q. Is x <= f(p) ? oassume: the send rate is a stationary ergodic process

19
19 Conditions for Conservativeness In practice: othe conditions are true, or almost othe result explains overly conservativeness

20
20 Is Negative or Slightly Positive ? Internet LAN to LAN EPFL sender Internet LAN to cable-modem at EPFL Lab

21
21 Throughput-Drop Puzzle Empirical indications: TFRC looses throughput for large loss-event rates E.g. Bansal et al (ACM SIGCOMM 2001): … in return to for smoother transmission rates, slowly-responsive algorithms lose throughput to faster ones (like TCP) under dynamic network conditions. Why ? L=2 4 8 16 PFTK-simplified Cause: convexity of 1/f(1/x) PFTK SQRT

22
22 What Causes Excessive Conservativeness ? Palm inversion: Throughput: May make the control conservative ? !

23
23 What Causes Excessive Conservativeness ? (Contd) o the overshoot bounded by a function of p and o 1/f(1/x) is assumed to be convex, thus, it is above its tangents o take the tangent at 1/p

24
24 Conclusion 1. Breakdown the TCP-friendliness into sub-conditions, study the sub-conditions separately 2. TCP-friendliness is difficult to verify 3. Conservativeness is easier

Similar presentations

OK

-Mayukh, clemson university1 Project Overview Study of Tfrc Verification, Analysis and Development Verification : Experiments. Analysis : Check for short.

-Mayukh, clemson university1 Project Overview Study of Tfrc Verification, Analysis and Development Verification : Experiments. Analysis : Check for short.

© 2018 SlidePlayer.com Inc.

All rights reserved.

To make this website work, we log user data and share it with processors. To use this website, you must agree to our Privacy Policy, including cookie policy.

Ads by Google