1

EE 122:TCP Congestion Control

Ion Stoica

TAs: Junda Liu, DK Moon, David Zats

http://inst.eecs.berkeley.edu/~ee122/(Materials with thanks to Vern Paxson, Jennifer Rexford,

and colleagues at UC Berkeley)

2

Goals of Today’s Lecture

Principles of congestion control Learning that congestion is occurring Adapting to alleviate the congestion

TCP congestion control Additive-increase, multiplicative-decrease (AIMD) How to begin transmitting: Slow Start

3

What We Know

We know: How to process packets in a switch How to route packets in the network How to send packets reliably

We don’t know: How fast to send

4

It’s Not Just The Sender & Receiver

Flow control keeps one fast sender from overwhelming a slow receiver

Congestion control keeps a set of senders from overloading the network

Three congestion control problems: Adjusting to bottleneck bandwidth

Without any a priori knowledge Could be a Gbps link; could be a modem

Adjusting to variations in bandwidth Sharing bandwidth between flows

5

Congestion is Unavoidable Two packets arrive at the same time

The node can only transmit one … and either buffers or drops the other

If many packets arrive in a short period of time The node cannot keep up with the arriving traffic … and the buffer may eventually overflow

6

Congestion Collapse

Definition: Increase in network load results in a decrease of useful work done

Due to: Undelivered packets

Packets consume resources and are dropped later in network

Spurious retransmissions of packets still in flight Unnecessary retransmissions lead to more load! Pouring gasoline on a fire

Mid-1980s: Internet grinds to a halt Until Jacobson/Karels (Berkeley!) devise TCP congestion

control

7

View from a Single Flow

Knee – point after which Throughput increases very

slowly Delay increases quickly

Cliff – point after which Throughput starts to

decrease very fast to zero (congestion collapse)

Delay approaches infinity

Load

Load

Th

rou

ghp

ut

De

lay

knee cliff

congestioncollapse

packetloss

8

General Approaches

Send without care Many packet drops

(1) Reservations Pre-arrange bandwidth allocations Requires negotiation before sending packets Low utilization

(2) Pricing Don’t drop packets for the high-bidders Requires payment model

9

General Approaches (cont’d)

(3) Dynamic Adjustment Probe network to test level of congestion Speed up when no congestion Slow down when congestion Suboptimal, messy dynamics, simple to implement

All three techniques have their place But for generic Internet usage, dynamic adjustment is

the most appropriate Due to pricing structure, traffic characteristics, and

good citizenship

10

TCP Congestion Control

TCP connection has window Controls number of unacknowledged packets

Sending rate: ~Window/RTT

Vary window size to control sending rate

11

Sizing the Windows

cwnd (Congestion Windows) How many bytes can be sent without

overflowing routers Computed by congestion control algorithm

AdvertisedWindow How many bytes can be sent without

overflowing the sender Determined by the receiver

12

EffectiveWindow

Limits how much data can be in transit Implemented as # of bytes Described as # packets (segments) in this

lecture

EffectiveWindow = MaxWindow – (LastByteSent – LastByteAcked)

MaxWindow = min(cwnd, AdvertisedWindow)

LastByteAckedLastByteSent

sequence number increases

MaxWindow

EffectiveWindow

13

Two Basic Components

Detecting congestion

Rate adjustment algorithm Depends on congestion or not Three subproblems within adjustment problem

Finding fixed bandwidth Adjusting to bandwidth variations Sharing bandwidth

14

Detecting Congestion

Packet dropping is best sign of congestion Delay-based methods are hard and risky

How do you detect packet drops? ACKs TCP uses ACKs to signal receipt of data ACK denotes last contiguous byte received

Actually, ACKs indicate next segment expected

Two signs of packet drops No ACK after certain time interval: time-out Several duplicate ACKs (ignore for now)

15

Rate Adjustment

Basic structure: Upon receipt of ACK (of new data): increase rate Upon detection of loss: decrease rate

But what increase/decrease functions should we use? Depends on what problem we are solving

16

Problem #1: Single Flow, Fixed BW

Want to get a first-order estimate of the available bandwidth Assume bandwidth is fixed Ignore presence of other flows

Want to start slow, but rapidly increase rate until packet drop occurs (“slow-start”)

Adjustment: cwnd initially set to 1 cwnd++ upon receipt of ACK

17

Slow-Start cwnd increases exponentially: cwnd doubles every

time a full cwnd of packets has been sent Each ACK releases two packets Slow-start is called “slow” because of starting point

segment 1cwnd = 1

cwnd = 2 segment 2segment 3

cwnd = 4 segment 4

segment 5segment 6segment 7

cwnd = 8

cwnd = 3

18

5 Minute Break

Questions Before We Proceed?

19

Problems with Slow-Start

Slow-start can result in many losses Roughly the size of cwnd ~ BW*RTT

Example: At some point, cwnd is enough to fill “pipe” After another RTT, cwnd is double its previous value All the excess packets are dropped!

Need a more gentle adjustment algorithm once have rough estimate of bandwidth

20

Problem #2: Single Flow, Varying BW

Want to be able to track available bandwidth, oscillating around its current value

Possible variations: (in terms of RTTs) Multiplicative increase or decrease: cwnd a*cwnd Additive increase or decrease: cwnd cwnd + b

Four alternatives: AIAD: gentle increase, gentle decrease AIMD: gentle increase, drastic decrease MIAD: drastic increase, gentle decrease (too many losses) MIMD: drastic increase and decrease

21

Problem #3: Multiple Flows

Want steady state to be “fair”

Many notions of fairness, but here all we require is that two identical flows end up with the same bandwidth

This eliminates MIMD and AIAD

AIMD is the only remaining solution!

22

Buffer and Window Dynamics

No congestion x increases by one packet/RTT every RTT Congestion decrease x by factor 2

A BC = 50 pkts/RTT

0

10

20

30

40

50

60

1 28 55 82 109

136

163

190

217

244

271

298

325

352

379

406

433

460

487

Backlog in router (pkts)Congested if > 20

Rate (pkts/RTT)

x

23

AIMD Sharing DynamicsA Bx1

D E

0

10

20

30

40

50

60

1 28 55 82 109

136

163

190

217

244

271

298

325

352

379

406

433

460

487

No congestion rate increases by one packet/RTT every RTT Congestion decrease rate by factor 2

Rates equalize fair share

x2

24

AIAD Sharing Dynamics

A Bx1

D E No congestion x increases by one packet/RTT every RTT Congestion decrease x by 1

0

10

20

30

40

50

60

1 28 55 82 109

136

163

190

217

244

271

298

325

352

379

406

433

460

487

x2

25

Efficient Allocation: Challenges of Congestion Control

Too slow Fail to take advantage of

available bandwidth underload

Too fast Overshoot knee overload,

high delay, loss Everyone’s doing it

May all under/over shoot large oscillations

Optimal: xi=Xgoal

Efficiency = 1 - distance from efficiency line

User 1: x1U

ser

2: x

2

Efficiencyline

2 user example

overload

underload

26

Example

User 1: x1

Use

r 2:

x2

fairnessline

efficiencyline

1

1

Total bandwidth 1

Inefficient: x1+x2=0.7

(0.2, 0.5)

Congested: x1+x2=1.2

(0.7, 0.5)

Efficient: x1+x2=1Not fair

(0.7, 0.3)

Efficient: x1+x2=1Fair

(0.5, 0.5)

27

MIAD

User 1: x1

Use

r 2:

x2

fairnessline

efficiencyline

(x1h,x2h)

(x1h-aD,x2h-aD)

(bI(x1h-aD), bI(x2h-aD)) Increase: x*bI

Decrease: x - aD

Does not converge to fairness

Does not converges to efficiency

28

AIAD

User 1: x1

Use

r 2:

x2

fairnessline

efficiencyline

(x1h,x2h)

(x1h-aD,x2h-aD)

(x1h-aD+aI),x2h-aD+aI)) Increase: x + aI

Decrease: x - aD

Does not converge to fairness

Does not converge to efficiency

29

MIMD

User 1: x1

Use

r 2:

x2

fairnessline

efficiencyline

(x1h,x2h)

(bdx1h,bdx2h)

(bIbDx1h,bIbDx2h)

Increase: x*bI

Decrease: x*bD

Does not converge to fairness

Converges to efficiency iff

10

1

D

I

b

b

30

(bDx1h+aI,bDx2h+aI)

AIMD

User 1: x1

Use

r 2:

x2

fairnessline

efficiencyline

(x1h,x2h)

(bDx1h,bDx2h)

Increase: x+aD

Decrease: x*bD

Converges to fairness Converges to

efficiency Increments smaller

as fairness increases

31

Implementing AIMD

After each ACK Increment cwnd by 1/cwnd (cwnd += 1/cwnd) As a result, cwnd is increased by one only if all

segments in a cwnd have been acknowledged

But need to decide when to leave slow-start and enter AIMD Use ssthresh variable

32

Slow Start/AIMD Pseudocode

Initially:cwnd = 1;ssthresh = infinite;

New ack received:if (cwnd < ssthresh) /* Slow Start*/ cwnd = cwnd + 1;else /* Congestion Avoidance */ cwnd = cwnd + 1/cwnd;

Timeout:/* Multiplicative decrease */ssthresh = cwnd/2;cwnd = 1;

33

The big picture (with timeouts)

Time

cwnd

Timeout

SlowStart

AIMD

ssthresh

Timeout

SlowStart

SlowStart

AIMD

34

Summary

Congestion is inevitable Internet does not reserve resources in advance TCP actively tries to grab capacity

Congestion control critical for avoiding collapse AIMD: Additive Increase, Multiplicative Decrease Congestion detected via packet loss (fail-safe) Slow start to find initial sending rate & to restart after

timeout Next class

Advanced congestion control