Chapter 3

Transport Layer

A note on the use of these Powerpoint slides:

We’re making these slides freely available to all (faculty, students, readers).
They’re in PowerPoint form so you see the animations; and can add, modify,
and delete slides (including this one) and slide content to suit your needs.

Computer
They obviously represent a lot of work on our part. In return for use, we only
ask the following:

 If you use these slides (e.g., in a class) that you mention their source
(after all, we’d like people to use our book!)
Networking: A Top
 If you post any slides on a www site, that you note that they are adapted
from (or perhaps identical to) our slides, and note our copyright of this
Down Approach
material.
7th edition
Thanks and enjoy! JFK/KWR
Jim Kurose, Keith Ross
All material copyright 1996-2016 Pearson/Addison Wesley
J.F Kurose and K.W. Ross, All Rights Reserved April 2016
Transport Layer 2-1
Chapter 3: Transport Layer
our goals:
 understand principles  learn about Internet
behind transport
layer services:
• multiplexing,
demultiplexing
• reliable data transfer
• flow control
• congestion control
transport layer protocols:
• UDP: connectionless
transport
• TCP: connection-oriented
reliable transport
• TCP congestion control
Transport Layer 3-2
Chapter 3 outline
3.1 transport-layer 3.5 connection-oriented
services transport: TCP
3.2 multiplexing and • segment structure
demultiplexing • reliable data transfer
3.3 connectionless • flow control
transport: UDP • connection management
3.4 principles of reliable 3.6 principles of congestion
data transfer control
3.7 TCP congestion control

Transport Layer 3-3

Transport services and protocols
application
transport
 provide logical communication network
data link
between app processes physical

running on different hosts

 transport protocols run in
end systems
• send side: breaks app
messages into segments,
passes to network layer
• rcv side: reassembles application
segments into messages, transport
network
passes to app layer data link
physical

 more than one transport

protocol available to apps
• Internet: TCP and UDP
Transport Layer 3-4
Transport vs. network layer
 network layer: logical household analogy:
communication
between hosts 12 kids in Ann’s house sending
letters to 12 kids in Bill’s
 transport layer: house:
logical  hosts = houses
communication  processes = kids
between processes  app messages = letters in
envelopes
• relies on, enhances,  transport protocol = Ann
network layer and Bill who demux to in-
services house siblings
 network-layer protocol =
postal service

Transport Layer 3-5

Internet transport-layer protocols
application
 reliable, in-order transport
network

delivery (TCP) data link

physical
network

• congestion control network

data link
data link
physical
physical
• flow control network
data link

• connection setup physical

network

 unreliable, unordered data link

physical

delivery: UDP network

data link
physical
• no-frills extension of network
data link application
“best-effort” IP physical
network
data link
transport
network
data link
 services not available: physical
physical

• delay guarantees
• bandwidth guarantees

Transport Layer 3-6

Chapter 3 outline
3.1 transport-layer 3.5 connection-oriented
services transport: TCP
3.2 multiplexing and • segment structure
demultiplexing • reliable data transfer
3.3 connectionless • flow control
transport: UDP • connection management
3.4 principles of reliable 3.6 principles of congestion
data transfer control
3.7 TCP congestion control

Transport Layer 3-7

Multiplexing/demultiplexing
multiplexing at sender:
handle data from multiple
sockets, add transport header
(later used for demultiplexing)
demultiplexing at receiver:
use header info to deliver
received segments to correct
socket

application

application P1 P2 application socket

P3 transport P4
process
transport network transport
network link network
link physical link
physical physical

Transport Layer 3-8

How demultiplexing works
 host receives IP datagrams 32 bits
• each datagram has source IP source port # dest port #
address, destination IP
address
other header fields
• each datagram carries one
transport-layer segment
• each segment has source, application
destination port number data
 host uses IP addresses & (payload)
port numbers to direct
segment to appropriate
TCP/UDP segment format
socket

Transport Layer 3-9

 Connectionless demultiplexing
 recall: created socket has
host-local port #:
DatagramSocket mySocket1
= new DatagramSocket(12534);
 when host receives UDP
segment:
• checks destination port #
in segment
• directs UDP segment to
socket with that port #
 recall: when creating
datagram to send into UDP
socket, must specify
• destination IP address
• destination port #
IP datagrams with same
dest. port #, but different
source IP addresses
and/or source port
numbers will be directed
to same socket at dest
Transport Layer 3-10
Connectionless demux: example
DatagramSocket
DatagramSocket serverSocket = new
DatagramSocket DatagramSocket
mySocket2 = new mySocket1 = new
DatagramSocket (6428); DatagramSocket
(9157); application
(5775);
application application
P1
P3 P4
transport
transport transport
network
network link network
link physical link
physical physical

source port: 6428 source port: ?

dest port: 9157 dest port: ?

source port: 9157 source port: ?

dest port: 6428 dest port: ?
Transport Layer 3-11
 Connection-oriented demux
 TCP socket identified  server host may support
by 4-tuple: many simultaneous TCP
• source IP address sockets:
• source port number • each socket identified by
• dest IP address its own 4-tuple
• dest port number  web servers have
 demux: receiver uses all different sockets for
four values to direct each connecting client
segment to appropriate • non-persistent HTTP will
socket have different socket for
each request

Transport Layer 3-12

 Connection-oriented demux: example

application
application P4 P5 P6 application
P3 P2 P3
transport
transport transport
network
network link network
link physical link
physical server: IP physical
address B

host: IP source IP,port: B,80 host: IP

address A dest IP,port: A,9157 source IP,port: C,5775 address C
dest IP,port: B,80
source IP,port: A,9157
dest IP, port: B,80
source IP,port: C,9157
dest IP,port: B,80
three segments, all destined to IP address: B,
dest port: 80 are demultiplexed to different sockets Transport Layer 3-13
 Connection-oriented demux: example
threaded server
application
application application
P4
P3 P2 P3
transport
transport transport
network
network link network
link physical link
physical server: IP physical
address B

host: IP source IP,port: B,80 host: IP

address A dest IP,port: A,9157 source IP,port: C,5775 address C
dest IP,port: B,80
source IP,port: A,9157
dest IP, port: B,80
source IP,port: C,9157
dest IP,port: B,80

Transport Layer 3-14

Chapter 3 outline
3.1 transport-layer 3.5 connection-oriented
services transport: TCP
3.2 multiplexing and • segment structure
demultiplexing • reliable data transfer
3.3 connectionless • flow control
transport: UDP • connection management
3.4 principles of reliable 3.6 principles of congestion
data transfer control
3.7 TCP congestion control

Transport Layer 3-15

UDP: User Datagram Protocol [RFC 768]
 “no frills,” “bare bones”  UDP use:
Internet transport  streaming multimedia
protocol apps (loss tolerant, rate
 “best effort” service, UDP sensitive)
segments may be:  DNS
• lost  SNMP
• delivered out-of-order  reliable transfer over
to app
UDP:
 connectionless:
 add reliability at
• no handshaking application layer
between UDP sender,
receiver  application-specific error
recovery!
• each UDP segment
handled independently
of others
Transport Layer 3-16
UDP: segment header
length, in bytes of
32 bits UDP segment,
source port # dest port # including header

length checksum
why is there a UDP?
 no connection
application establishment (which can
data add delay)
(payload)  simple: no connection
state at sender, receiver
 small header size
UDP segment format  no congestion control:
UDP can blast away as fast
as desired

Transport Layer 3-17

UDP checksum
Goal: detect “errors” (e.g., flipped bits) in transmitted
segment
sender: receiver:
 treat segment contents,  compute checksum of
including header fields, received segment
as sequence of 16-bit  check if computed checksum
integers
equals checksum field value:
 checksum: addition
(one’s complement sum) • NO - error detected
of segment contents • YES - no error detected.
 sender puts checksum But maybe errors
value into UDP checksum nonetheless? More later
field ….

Transport Layer 3-18

Internet checksum: example
example: add two 16-bit integers
1 1 1 1 0 0 1 1 0 0 1 1 0 0 1 1 0
1 1 1 0 1 0 1 0 1 0 1 0 1 0 1 0 1

wraparound 1 1 0 1 1 1 0 1 1 1 0 1 1 1 0 1 1

sum 1 1 0 1 1 1 0 1 1 1 0 1 1 1 1 0 0
checksum 1 0 1 0 0 0 1 0 0 0 1 0 0 0 0 1 1

Note: when adding numbers, a carryout from the most

significant bit needs to be added to the result

* Check out the online interactive exercises for more

examples: http://gaia.cs.umass.edu/kurose_ross/interactive/ Transport Layer 3-19
UDP, TCP and IP checksums
 All work using the same algorithm
 An example can be found here:
• http://www.thegeekstuff.com/2012/05/ip-
header-checksum/

Transport Layer 3-20

Pseudo header
 UDP checksums header, data, and pseudo-
header:

Transport Layer 3-21

Why pseudo header?
 historical  not really useful

Transport Layer 3-22

binary numbers
 http://l3d.cs.colorado.edu/courses/CSCI120
0-96/binary.html
 http://www.mathsisfun.com/binary-number-
system.html
 http://www.wikihow.com/Add-Binary-
Numbers

Transport Layer 3-23

Chapter 3 outline
3.1 transport-layer 3.5 connection-oriented
services transport: TCP
3.2 multiplexing and • segment structure
demultiplexing • reliable data transfer
3.3 connectionless • flow control
transport: UDP • connection management
3.4 principles of reliable 3.6 principles of congestion
data transfer control
3.7 TCP congestion control

Transport Layer 3-24

Principles of reliable data transfer
 important in application, transport, link layers
• top-10 list of important networking topics!

 characteristics of unreliable channel will determine

complexity of reliable data transfer protocol (rdt)
Transport Layer 3-25
Principles of reliable data transfer
 important in application, transport, link layers
• top-10 list of important networking topics!

 characteristics of unreliable channel will determine

complexity of reliable data transfer protocol (rdt)
Transport Layer 3-26
Principles of reliable data transfer
 important in application, transport, link layers
• top-10 list of important networking topics!

 characteristics of unreliable channel will determine

complexity of reliable data transfer protocol (rdt)
Transport Layer 3-27
Reliable data transfer: getting started
rdt_send(): called from above, deliver_data(): called by
(e.g., by app.). Passed data to rdt to deliver data to upper
deliver to receiver upper layer

send receive
side side

udt_send(): called by rdt, rdt_rcv(): called when packet

to transfer packet over arrives on rcv-side of channel
unreliable channel to receiver

Transport Layer 3-28

Reliable data transfer: getting started
we’ll:
 incrementally develop sender, receiver sides of
reliable data transfer protocol (rdt)
 consider only unidirectional data transfer
• but control info will flow on both directions!
 use finite state machines (FSM) to specify sender,
receiver
event causing state transition
actions taken on state transition
state: when in this
“state” next state state state
uniquely determined 1 event
by next event 2
actions

Transport Layer 3-29

rdt1.0: reliable transfer over a reliable channel
 underlying channel perfectly reliable
• no bit errors
• no loss of packets
 separate FSMs for sender, receiver:
• sender sends data into underlying channel
• receiver reads data from underlying channel

Wait for rdt_send(data) Wait for rdt_rcv(packet)

call from call from extract (packet,data)
above packet = make_pkt(data) below deliver_data(data)
udt_send(packet)

sender receiver

Transport Layer 3-30

rdt2.0: channel with bit errors
 underlying channel may flip bits in packet
• checksum to detect bit errors
 the question: how to recover from errors:
• acknowledgements (ACKs): receiver explicitly tells sender
that pkt received OK
• negative acknowledgements (NAKs): receiver explicitly tells
sender that pkt had errors
• sender
Howretransmits
do humanspkt on receipt from
recover of NAK“errors”
 new mechanisms in rdt2.0 (beyond rdt1.0):
• error detection
during conversation?
• receiver feedback: control msgs (ACK,NAK) rcvr-
>sender

Transport Layer 3-31

rdt2.0: channel with bit errors
 underlying channel may flip bits in packet
• checksum to detect bit errors
 the question: how to recover from errors:
• acknowledgements (ACKs): receiver explicitly tells sender
that pkt received OK
• negative acknowledgements (NAKs): receiver explicitly tells
sender that pkt had errors
• sender retransmits pkt on receipt of NAK
 new mechanisms in rdt2.0 (beyond rdt1.0):
• error detection
• feedback: control msgs (ACK,NAK) from receiver to
sender

Transport Layer 3-32

rdt2.0: FSM specification
rdt_send(data)
sndpkt = make_pkt(data, checksum) receiver
udt_send(sndpkt)
rdt_rcv(rcvpkt) &&
isNAK(rcvpkt)
Wait for Wait for rdt_rcv(rcvpkt) &&
call from ACK or udt_send(sndpkt) corrupt(rcvpkt)
above NAK
udt_send(NAK)

rdt_rcv(rcvpkt) && isACK(rcvpkt)

Wait for
L
call from
sender below

rdt_rcv(rcvpkt) &&
notcorrupt(rcvpkt)
extract(rcvpkt,data)
deliver_data(data)
udt_send(ACK)

Transport Layer 3-33

rdt2.0: operation with no errors
rdt_send(data)
snkpkt = make_pkt(data, checksum)
udt_send(sndpkt)
rdt_rcv(rcvpkt) &&
isNAK(rcvpkt)
Wait for Wait for rdt_rcv(rcvpkt) &&
call from ACK or udt_send(sndpkt) corrupt(rcvpkt)
above NAK
udt_send(NAK)

rdt_rcv(rcvpkt) && isACK(rcvpkt)

Wait for
L call from
below

rdt_rcv(rcvpkt) &&
notcorrupt(rcvpkt)
extract(rcvpkt,data)
deliver_data(data)
udt_send(ACK)

Transport Layer 3-34

rdt2.0: error scenario
rdt_send(data)
snkpkt = make_pkt(data, checksum)
udt_send(sndpkt)
rdt_rcv(rcvpkt) &&
isNAK(rcvpkt)
Wait for Wait for rdt_rcv(rcvpkt) &&
call from ACK or udt_send(sndpkt) corrupt(rcvpkt)
above NAK
udt_send(NAK)

rdt_rcv(rcvpkt) && isACK(rcvpkt)

Wait for
L call from
below

rdt_rcv(rcvpkt) &&
notcorrupt(rcvpkt)
extract(rcvpkt,data)
deliver_data(data)
udt_send(ACK)

Transport Layer 3-35

rdt2.0 has a fatal flaw!
what happens if handling duplicates:
ACK/NAK corrupted?  sender retransmits
 sender doesn’t know current pkt if ACK/NAK
what happened at corrupted
receiver!
 sender adds sequence
 can’t just retransmit: number to each pkt
possible duplicate
 receiver discards (doesn’t
deliver up) duplicate pkt
stop and wait
sender sends one packet,
then waits for receiver
response

Transport Layer 3-36

rdt2.1: sender, handles garbled ACK/NAKs
rdt_send(data)
sndpkt = make_pkt(0, data, checksum)
udt_send(sndpkt) rdt_rcv(rcvpkt) &&
( corrupt(rcvpkt) ||
Wait for Wait for
ACK or
isNAK(rcvpkt) )
call 0 from
NAK 0 udt_send(sndpkt)
above
rdt_rcv(rcvpkt)
&& notcorrupt(rcvpkt) rdt_rcv(rcvpkt)
&& isACK(rcvpkt) && notcorrupt(rcvpkt)
&& isACK(rcvpkt)
L
L
Wait for Wait for
ACK or call 1 from
rdt_rcv(rcvpkt) && NAK 1 above
( corrupt(rcvpkt) ||
isNAK(rcvpkt) ) rdt_send(data)

udt_send(sndpkt) sndpkt = make_pkt(1, data, checksum)

udt_send(sndpkt)

Transport Layer 3-37

 rdt2.1: receiver, handles garbled ACK/NAKs
rdt_rcv(rcvpkt) && notcorrupt(rcvpkt)
&& has_seq0(rcvpkt)
extract(rcvpkt,data)
deliver_data(data)
sndpkt = make_pkt(ACK, chksum)
udt_send(sndpkt)
rdt_rcv(rcvpkt) && (corrupt(rcvpkt) rdt_rcv(rcvpkt) && (corrupt(rcvpkt)
sndpkt = make_pkt(NAK, chksum) sndpkt = make_pkt(NAK, chksum)
udt_send(sndpkt) udt_send(sndpkt)
Wait for Wait for
rdt_rcv(rcvpkt) && 0 from 1 from rdt_rcv(rcvpkt) &&
not corrupt(rcvpkt) && below below not corrupt(rcvpkt) &&
has_seq1(rcvpkt) has_seq0(rcvpkt)
sndpkt = make_pkt(ACK, chksum) sndpkt = make_pkt(ACK, chksum)
udt_send(sndpkt) udt_send(sndpkt)
rdt_rcv(rcvpkt) && notcorrupt(rcvpkt)
&& has_seq1(rcvpkt)

extract(rcvpkt,data)
deliver_data(data)
sndpkt = make_pkt(ACK, chksum)
udt_send(sndpkt)

Transport Layer 3-38

rdt2.1: discussion
sender: receiver:
 seq # added to pkt  must check if received
 two seq. #’s (0,1) will packet is duplicate
suffice. Why? • state indicates whether
0 or 1 is expected pkt
 must check if received seq #
ACK/NAK corrupted
 note: receiver can not
 twice as many states know if its last
• state must ACK/NAK received
“remember” whether OK at sender
“expected” pkt should
have seq # of 0 or 1

Transport Layer 3-39

rdt2.2: a NAK-free protocol
 same functionality as rdt2.1, using ACKs only
 instead of NAK, receiver sends ACK for last pkt
received OK
• receiver must explicitly include seq # of pkt being ACKed
 duplicate ACK at sender results in same action as
NAK: retransmit current pkt

Transport Layer 3-40

 rdt2.2: sender, receiver fragments
rdt_send(data)
sndpkt = make_pkt(0, data, checksum)
udt_send(sndpkt)
rdt_rcv(rcvpkt) &&
( corrupt(rcvpkt) ||
Wait for Wait for
ACK isACK(rcvpkt,1) )
call 0 from
above 0 udt_send(sndpkt)
sender FSM
fragment rdt_rcv(rcvpkt)
&& notcorrupt(rcvpkt)
rdt_rcv(rcvpkt) && && isACK(rcvpkt,0)
(corrupt(rcvpkt) || L
has_seq1(rcvpkt)) Wait for receiver FSM
0 from
udt_send(sndpkt) below fragment
rdt_rcv(rcvpkt) && notcorrupt(rcvpkt)
&& has_seq1(rcvpkt)
extract(rcvpkt,data)
deliver_data(data)
sndpkt = make_pkt(ACK1, chksum)
udt_send(sndpkt) Transport Layer 3-41
rdt3.0: channels with errors and loss

new assumption: approach: sender waits

underlying channel can “reasonable” amount of
also lose packets (data, time for ACK
ACKs)  retransmits if no ACK
• checksum, seq. #, received in this time
ACKs, retransmissions  if pkt (or ACK) just delayed
(not lost):
will be of help … but
not enough • retransmission will be
duplicate, but seq. #’s
already handles this
• receiver must specify seq
# of pkt being ACKed
 requires countdown timer

Transport Layer 3-42

rdt3.0 sender
rdt_send(data)
rdt_rcv(rcvpkt) &&
sndpkt = make_pkt(0, data, checksum) ( corrupt(rcvpkt) ||
udt_send(sndpkt) isACK(rcvpkt,1) )
rdt_rcv(rcvpkt) start_timer L
L Wait for Wait
for timeout
call 0from
ACK0 udt_send(sndpkt)
above
start_timer
rdt_rcv(rcvpkt)
&& notcorrupt(rcvpkt) rdt_rcv(rcvpkt)
&& isACK(rcvpkt,1) && notcorrupt(rcvpkt)
stop_timer && isACK(rcvpkt,0)
stop_timer
Wait Wait for
timeout for call 1 from
udt_send(sndpkt) ACK1 above
start_timer rdt_rcv(rcvpkt)
rdt_send(data) L
rdt_rcv(rcvpkt) &&
( corrupt(rcvpkt) || sndpkt = make_pkt(1, data, checksum)
isACK(rcvpkt,0) ) udt_send(sndpkt)
start_timer
L

Transport Layer 3-43

 rdt3.0 in action
sender receiver sender receiver
send pkt0 pkt0 send pkt0 pkt0
rcv pkt0 rcv pkt0
ack0 send ack0 ack0 send ack0
rcv ack0 rcv ack0
send pkt1 pkt1 send pkt1 pkt1
rcv pkt1 X
ack1 send ack1 loss
rcv ack1
send pkt0 pkt0
rcv pkt0 timeout
ack0 send ack0 resend pkt1 pkt1
rcv pkt1
ack1 send ack1
rcv ack1
send pkt0 pkt0
(a) no loss rcv pkt0
ack0 send ack0

(b) packet loss

Transport Layer 3-44
 rdt3.0 in action
sender receiver
sender receiver send pkt0 pkt0
send pkt0 pkt0 rcv pkt0
send ack0
rcv pkt0 ack0
send ack0 rcv ack0
ack0 send pkt1 pkt1
rcv ack0 rcv pkt1
send pkt1 pkt1
send ack1
rcv pkt1 ack1
ack1 send ack1
X
loss timeout
resend pkt1 pkt1
rcv pkt1
timeout
resend pkt1 pkt1 rcv ack1 (detect duplicate)
rcv pkt1 send pkt0
pkt0
send ack1
(detect duplicate) ack1
ack1 send ack1 rcv ack1 rcv pkt0
rcv ack1 send pkt0
ack0 send ack0
send pkt0 pkt0 pkt0
rcv pkt0
rcv pkt0 ack0 (detect duplicate)
ack0 send ack0 send ack0

(c) ACK loss (d) premature timeout/ delayed ACK

Transport Layer 3-45

Performance of rdt3.0
 rdt3.0 is correct, but performance stinks
 e.g.: 1 Gbps link, 15 ms prop. delay, 8000 bit packet:
L 8000 bits
Dtrans = R = = 8 microsecs
109 bits/sec

 U sender: utilization – fraction of time sender busy sending

U L/R .008
sender = = = 0.00027
RTT + L / R 30.008

 if RTT=30 msec, 1KB pkt every 30 msec: 33kB/sec thruput

over 1 Gbps link
 network protocol limits use of physical resources!
Transport Layer 3-46
 rdt3.0: stop-and-wait operation
sender receiver
first packet bit transmitted, t = 0
last packet bit transmitted, t = L / R

first packet bit arrives

RTT last packet bit arrives, send ACK

ACK arrives, send next

packet, t = RTT + L / R

U L/R .008
sender = = = 0.00027
RTT + L / R 30.008

Transport Layer 3-47

Pipelined protocols
pipelining: sender allows multiple, “in-flight”, yet-
to-be-acknowledged pkts
• range of sequence numbers must be increased
• buffering at sender and/or receiver

 two generic forms of pipelined protocols: go-Back-N,

selective repeat
Transport Layer 3-48
 Pipelining: increased utilization
sender receiver
first packet bit transmitted, t = 0
last bit transmitted, t = L / R

first packet bit arrives

RTT last packet bit arrives, send ACK
last bit of 2nd packet arrives, send ACK
last bit of 3rd packet arrives, send ACK
ACK arrives, send next
packet, t = RTT + L / R
3-packet pipelining increases
utilization by a factor of 3!

U 3L / R .0024
sender = = = 0.00081
RTT + L / R 30.008

Transport Layer 3-49

Pipelined protocols: overview
Go-back-N: Selective Repeat:
 sender can have up to  sender can have up to N
N unacked packets in unack’ed packets in
pipeline pipeline
 receiver only sends  rcvr sends individual ack
cumulative ack for each packet
• doesn’t ack packet if
there’s a gap
 sender has timer for  sender maintains timer
oldest unacked packet for each unacked packet
• when timer expires, • when timer expires,
retransmit all unacked retransmit only that
packets unacked packet

Transport Layer 3-50

Go-Back-N: sender
 k-bit seq # in pkt header
 “window” of up to N, consecutive unack’ed pkts allowed

 ACK(n): ACKs all pkts up to, including seq # n - “cumulative

ACK”
• may receive duplicate ACKs (see receiver)
 timer for oldest in-flight pkt
 timeout(n): retransmit packet n and all higher seq # pkts in
window
Transport Layer 3-51
GBN: sender extended FSM
rdt_send(data)
if (nextseqnum < base+N) {
sndpkt[nextseqnum] = make_pkt(nextseqnum,data,chksum)
udt_send(sndpkt[nextseqnum])
if (base == nextseqnum)
start_timer
nextseqnum++
}
L else
refuse_data(data)
base=1
nextseqnum=1
timeout
start_timer
Wait
udt_send(sndpkt[base])
rdt_rcv(rcvpkt) udt_send(sndpkt[base+1])
&& corrupt(rcvpkt) …
udt_send(sndpkt[nextseqnum-1])
rdt_rcv(rcvpkt) &&
notcorrupt(rcvpkt)
base = getacknum(rcvpkt)+1
If (base == nextseqnum)
stop_timer
else
start_timer
Transport Layer 3-52
GBN: receiver extended FSM
default
udt_send(sndpkt) rdt_rcv(rcvpkt)
&& notcurrupt(rcvpkt)
L && hasseqnum(rcvpkt,expectedseqnum)
expectedseqnum=1 Wait extract(rcvpkt,data)
sndpkt = deliver_data(data)
make_pkt(expectedseqnum,ACK,chksum) sndpkt = make_pkt(expectedseqnum,ACK,chksum)
udt_send(sndpkt)
expectedseqnum++

ACK-only: always send ACK for correctly-received

pkt with highest in-order seq #
• may generate duplicate ACKs
• need only remember expectedseqnum
 out-of-order pkt:
• discard (don’t buffer): no receiver buffering!
• re-ACK pkt with highest in-order seq #
Transport Layer 3-53
 GBN in action
sender window (N=4) sender receiver
012345678 send pkt0
012345678 send pkt1
012345678 send pkt2 receive pkt0, send ack0
012345678 send pkt3 Xloss receive pkt1, send ack1
(wait)
receive pkt3, discard,
012345678 rcv ack0, send pkt4 (re)send ack1
012345678 rcv ack1, send pkt5 receive pkt4, discard,
(re)send ack1
ignore duplicate ACK receive pkt5, discard,
(re)send ack1
pkt 2 timeout
012345678 send pkt2
012345678 send pkt3
012345678 send pkt4 rcv pkt2, deliver, send ack2
012345678 send pkt5 rcv pkt3, deliver, send ack3
rcv pkt4, deliver, send ack4
rcv pkt5, deliver, send ack5

Transport Layer 3-54

Selective repeat
 receiver individually acknowledges all correctly
received pkts
• buffers pkts, as needed, for eventual in-order delivery
to upper layer
 sender only resends pkts for which ACK not
received
• sender timer for each unACKed pkt
 sender window
• N consecutive seq #’s
• limits seq #s of sent, unACKed pkts

Transport Layer 3-55

Selective repeat: sender, receiver windows

Transport Layer 3-56

Selective repeat
sender
data from above:
 if next available seq # in
window, send pkt
timeout(n):
 resend pkt n, restart timer
ACK(n) in [sendbase,sendbase+N]:
 mark pkt n as received
 if n smallest unACKed pkt,
advance window base to
next unACKed seq #
receiver
pkt n in [rcvbase, rcvbase+N-1]
 send ACK(n)
 out-of-order: buffer
 in-order: deliver (also
deliver buffered, in-order
pkts), advance window to
next not-yet-received pkt
pkt n in [rcvbase-N,rcvbase-1]
 ACK(n)
otherwise:
 ignore
Transport Layer 3-57
 Selective repeat in action
sender window (N=4) sender receiver
012345678 send pkt0
012345678 send pkt1
012345678 send pkt2 receive pkt0, send ack0
012345678 send pkt3 Xloss receive pkt1, send ack1
(wait)
receive pkt3, buffer,
012345678 rcv ack0, send pkt4 send ack3
012345678 rcv ack1, send pkt5 receive pkt4, buffer,
send ack4
record ack3 arrived receive pkt5, buffer,
send ack5
pkt 2 timeout
012345678 send pkt2
012345678 record ack4 arrived
012345678 rcv pkt2; deliver pkt2,
record ack5 arrived
012345678 pkt3, pkt4, pkt5; send ack2

Q: what happens when ack2 arrives?

Transport Layer 3-58

 sender window receiver window
Selective repeat: (after receipt) (after receipt)

dilemma 0123012 pkt0

pkt1
0123012 0123012
pkt2 0123012
example:
0123012
0123012
pkt3
 seq #’s: 0, 1, 2, 3
0123012
X
0123012
 window size=3 pkt0 will accept packet
with seq number 0
(a) no problem
 receiver sees no
difference in two receiver can’t see sender side.
scenarios! receiver behavior identical in both cases!
something’s (very) wrong!
 duplicate data
accepted as new in (b) 0123012 pkt0
0123012 pkt1 0123012
pkt2
Q: what relationship 0123012
X
0123012
0123012
between seq # size X
and window size to timeout
retransmit pkt0 X
avoid problem in (b)? 0123012 pkt0
will accept packet
with seq number 0
(b) oops!
Transport Layer 3-59
Chapter 3 outline
3.1 transport-layer 3.5 connection-oriented
services transport: TCP
3.2 multiplexing and • segment structure
demultiplexing • reliable data transfer
3.3 connectionless • flow control
transport: UDP • connection management
3.4 principles of reliable 3.6 principles of congestion
data transfer control
3.7 TCP congestion control

Transport Layer 3-60

TCP: Overview RFCs: 793,1122,1323, 2018, 2581

 point-to-point:  full duplex data:

• one sender, one receiver • bi-directional data flow
 reliable, in-order byte in same connection
steam: • MSS: maximum segment
size
• no “message
boundaries”  connection-oriented:
 pipelined: • handshaking (exchange
of control msgs) inits
• TCP congestion and sender, receiver state
flow control set window before data exchange
size
 flow controlled:
• sender will not
overwhelm receiver
Transport Layer 3-61
 TCP segment structure
32 bits
URG: urgent data counting
(generally not used) source port # dest port #
by bytes
sequence number of data
ACK: ACK #
valid acknowledgement number (not segments!)
head not
PSH: push data now len used
UAP R S F receive window
(generally not used) # bytes
checksum Urg data pointer
rcvr willing
RST, SYN, FIN: to accept
options (variable length)
connection estab
(setup, teardown
commands)
application
Internet data
checksum (variable length)
(as in UDP)

Transport Layer 3-62

TCP seq. numbers, ACKs
sequence numbers:
• byte stream “number” of
outgoing segment from sender
source port # dest port #
sequence number
acknowledgement number

first byte in segment’s rwnd

data
checksum urg pointer

window size
acknowledgements: N

• seq # of next byte

expected from other side sender sequence number space
• cumulative ACK
sent sent, not- usable not
Q: how receiver handles ACKed yet ACKed but not usable
out-of-order segments (“in-
flight”)
yet sent

• A: TCP spec doesn’t say, incoming segment to sender

- up to implementor source port # dest port #
sequence number
acknowledgement number
A rwnd
checksum urg pointer

Transport Layer 3-63

TCP seq. numbers, ACKs
Host A Host B

User
types
‘C’ Seq=42, ACK=79, data = ‘C’
host ACKs
receipt of
‘C’, echoes
Seq=79, ACK=43, data = ‘C’ back ‘C’
host ACKs
receipt
of echoed
‘C’ Seq=43, ACK=80

simple telnet scenario

Transport Layer 3-64

TCP round trip time, timeout
Q: how to set TCP Q: how to estimate RTT?
timeout value?  SampleRTT: measured
time from segment
 longer than RTT transmission until ACK
• but RTT varies receipt
 too short: premature • ignore retransmissions
timeout, unnecessary  SampleRTT will vary, want
retransmissions estimated RTT “smoother”
• average several recent
 too long: slow reaction measurements, not just
to segment loss current SampleRTT

Transport Layer 3-65

TCP round trip time, timeout
EstimatedRTT = (1- )*EstimatedRTT + *SampleRTT
 exponential weighted moving average
 influence of past sample decreases exponentially fast
 typical value:  = 0.125 RTT: gaia.cs.umass.edu to fantasia.eurecom.fr

350

RTT: gaia.cs.umass.edu to fantasia.eurecom.fr

RTT (milliseconds)

300

250
RTT (milliseconds)

200

sampleRTT
150

EstimatedRTT

100
1 8 15 22 29 36 43 50 57 64 71 78 85 92 99 106
time (seconnds)
time (seconds) Transport Layer 3-66
SampleRTT Estimated RTT
 TCP round trip time, timeout
 timeout interval: EstimatedRTT plus “safety margin”
• large variation in EstimatedRTT -> larger safety margin
 estimate SampleRTT deviation from EstimatedRTT:
DevRTT = (1-)*DevRTT +
*|SampleRTT-EstimatedRTT|
(typically,  = 0.25)

TimeoutInterval = EstimatedRTT + 4*DevRTT

estimated RTT “safety margin”

* Check out the online interactive exercises for more

examples: http://gaia.cs.umass.edu/kurose_ross/interactive/ Transport Layer 3-67
Chapter 3 outline
3.1 transport-layer 3.5 connection-oriented
services transport: TCP
3.2 multiplexing and • segment structure
demultiplexing • reliable data transfer
3.3 connectionless • flow control
transport: UDP • connection management
3.4 principles of reliable 3.6 principles of congestion
data transfer control
3.7 TCP congestion control

Transport Layer 3-68

TCP reliable data transfer
 TCP creates rdt service
on top of IP’s unreliable
service
• pipelined segments
• cumulative acks let’s initially consider
• single retransmission simplified TCP sender:
timer • ignore duplicate acks
 retransmissions • ignore flow control,
triggered by: congestion control
• timeout events
• duplicate acks

Transport Layer 3-69

TCP sender events:
data rcvd from app: timeout:
 create segment with  retransmit segment
seq # that caused timeout
 seq # is byte-stream  restart timer
number of first data ack rcvd:
byte in segment  if ack acknowledges
 start timer if not previously unacked
already running segments
• think of timer as for • update what is known
oldest unacked to be ACKed
segment
• start timer if there are
• expiration interval: still unacked segments
TimeOutInterval

Transport Layer 3-70

TCP sender (simplified)
data received from application above
create segment, seq. #: NextSeqNum
pass segment to IP (i.e., “send”)
NextSeqNum = NextSeqNum + length(data)
if (timer currently not running)
L start timer
NextSeqNum = InitialSeqNum wait
SendBase = InitialSeqNum for
event timeout
retransmit not-yet-acked segment
with smallest seq. #
start timer
ACK received, with ACK field value y
if (y > SendBase) {
SendBase = y
/* SendBase–1: last cumulatively ACKed byte */
if (there are currently not-yet-acked segments)
start timer
else stop timer
} Transport Layer 3-71
TCP: retransmission scenarios
Host A Host B Host A Host B

SendBase=92
Seq=92, 8 bytes of data Seq=92, 8 bytes of data

Seq=100, 20 bytes of data

timeout

timeout
ACK=100
X
ACK=100
ACK=120

Seq=92, 8 bytes of data Seq=92, 8

SendBase=100 bytes of data
SendBase=120
ACK=100
ACK=120

SendBase=120

lost ACK scenario premature timeout

Transport Layer 3-72
TCP: retransmission scenarios
Host A Host B

Seq=92, 8 bytes of data

Seq=100, 20 bytes of data

timeout

ACK=100
X
ACK=120

Seq=120, 15 bytes of data

cumulative ACK
Transport Layer 3-73
TCP ACK generation [RFC 1122, RFC 2581]

event at receiver TCP receiver action

arrival of in-order segment with delayed ACK. Wait up to 500ms
expected seq #. All data up to for next segment. If no next segment,
expected seq # already ACKed send ACK

arrival of in-order segment with immediately send single cumulative

expected seq #. One other ACK, ACKing both in-order segments
segment has ACK pending

arrival of out-of-order segment immediately send duplicate ACK,

higher-than-expect seq. # . indicating seq. # of next expected byte
Gap detected

arrival of segment that immediate send ACK, provided that

partially or completely fills gap segment starts at lower end of gap

Transport Layer 3-74

TCP fast retransmit
 time-out period often
relatively long:
• long delay before
resending lost packet
 detect lost segments
via duplicate ACKs.
• sender often sends
many segments back-
to-back
• if segment is lost, there
will likely be many
duplicate ACKs.
TCP fast retransmit
if sender receives 3
ACKs for same data
(“triple
(“triple duplicate
duplicate ACKs”),
ACKs”),
resend unacked
segment with smallest
seq #
 likely that unacked
segment lost, so don’t
wait for timeout
Transport Layer 3-75
TCP fast retransmit
Host A Host B

Seq=92, 8 bytes of data

Seq=100, 20 bytes of data
X

ACK=100
timeout

ACK=100
ACK=100
ACK=100
Seq=100, 20 bytes of data

fast retransmit after sender

receipt of triple duplicate ACK
Transport Layer 3-76
(so 4th ack)
Quiz
 What is the diff between GBN and SR?
 When does TCP retransmit?
and how much will it retransmit?

Transport Layer 3-77

Quiz
 how is the timeout interval in tcp
calculated?
 what is delayed ack?
 what is fast retransmit?

Transport Layer 3-78

Chapter 3 outline
3.1 transport-layer 3.5 connection-oriented
services transport: TCP
3.2 multiplexing and • segment structure
demultiplexing • reliable data transfer
3.3 connectionless • flow control
transport: UDP • connection management
3.4 principles of reliable 3.6 principles of congestion
data transfer control
3.7 TCP congestion control

Transport Layer 3-79

TCP flow control
application
application may process
remove data from application
TCP socket buffers ….
TCP socket OS
receiver buffers
… slower than TCP
receiver is delivering
(sender is sending) TCP
code

IP
flow control code
receiver controls sender, so
sender won’t overflow
receiver’s buffer by transmitting from sender
too much, too fast
receiver protocol stack

Transport Layer 3-80

TCP flow control
 receiver “advertises” free
buffer space by including to application process
rwnd value in TCP header
of receiver-to-sender
segments RcvBuffer buffered data
• RcvBuffer size set via
socket options (typical default rwnd free buffer space
is 4096 bytes)
• many operating systems
autoadjust RcvBuffer TCP segment payloads
 sender limits amount of
unacked (“in-flight”) data to receiver-side buffering
receiver’s rwnd value
 guarantees receive buffer
will not overflow
Transport Layer 3-81
Chapter 3 outline
3.1 transport-layer 3.5 connection-oriented
services transport: TCP
3.2 multiplexing and • segment structure
demultiplexing • reliable data transfer
3.3 connectionless • flow control
transport: UDP • connection management
3.4 principles of reliable 3.6 principles of congestion
data transfer control
3.7 TCP congestion control

Transport Layer 3-82

Connection Management
before exchanging data, sender/receiver “handshake”:
 agree to establish connection (each knowing the other willing
to establish connection)
 agree on connection parameters

application application

connection state: ESTAB connection state: ESTAB

connection variables:
seq # client-to-server
server-to-client
rcvBuffer size
at server,client
connection Variables:
seq # client-to-server
server-to-client
rcvBuffer size
at server,client

network network

Socket clientSocket = Socket connectionSocket =

newSocket("hostname","port welcomeSocket.accept();
number");

Transport Layer 3-83

Agreeing to establish a connection

2-way handshake:
Q: will 2-way handshake
always work in
network?
Let’s talk
ESTAB  variable delays
OK
ESTAB  retransmitted messages (e.g.
req_conn(x)) due to
message loss
 message reordering
choose x
req_conn(x)
 can’t “see” other side
ESTAB
acc_conn(x)
ESTAB

Transport Layer 3-84

Agreeing to establish a connection
2-way handshake failure scenarios:

choose x choose x
req_conn(x) req_conn(x)
ESTAB ESTAB
retransmit acc_conn(x) retransmit acc_conn(x)
req_conn(x) req_conn(x)

ESTAB ESTAB
data(x+1) accept
req_conn(x)
retransmit data(x+1)
data(x+1)
connection connection
client x completes server x completes server
client
terminates forgets x terminates forgets x
req_conn(x)

ESTAB ESTAB
data(x+1) accept
half open connection! data(x+1)
(no client!)
Transport Layer 3-85
 TCP 3-way handshake

client state server state

LISTEN
choose init seq num, x
send TCP SYN msg
SYNSENT
received SYNACK(x)
ESTAB indicates server is live;
send ACK for SYNACK;
this segment may contain
client-to-server data
LISTEN
SYNbit=1, Seq=x
choose init seq num, y
send TCP SYNACK
msg, acking SYN SYN RCVD
SYNbit=1, Seq=y
ACKbit=1; ACKnum=x+1
ACKbit=1, ACKnum=y+1
received ACK(y)
indicates client is live
ESTAB

Transport Layer 3-86

TCP 3-way handshake: FSM

closed

Socket connectionSocket =
welcomeSocket.accept();

L Socket clientSocket =
SYN(x) newSocket("hostname","port
number");
SYNACK(seq=y,ACKnum=x+1)
create new socket for SYN(seq=x)
communication back to client listen

SYN SYN
rcvd sent

SYNACK(seq=y,ACKnum=x+1)
ESTAB ACK(ACKnum=y+1)
ACK(ACKnum=y+1)
L

Transport Layer 3-87

TCP: closing a connection
 client, server each close their side of connection
• send TCP segment with FIN bit = 1
 respond to received FIN with ACK
• on receiving FIN, ACK can be combined with own FIN
 simultaneous FIN exchanges can be handled

Transport Layer 3-88

TCP: closing a connection
client state server state
ESTAB ESTAB
clientSocket.close()
FIN_WAIT_1 can no longer FINbit=1, seq=x
send but can
receive data CLOSE_WAIT
ACKbit=1; ACKnum=x+1
can still
FIN_WAIT_2 wait for server send data
close

LAST_ACK
FINbit=1, seq=y
TIMED_WAIT can no longer
send data
ACKbit=1; ACKnum=y+1
timed wait
for 2*max CLOSED
segment lifetime

CLOSED

e.g. 30 sec, 1 minute, or 2 minutes Transport Layer 3-89

TCP Connection Management (cont)

TCP server
lifecycle

TCP client
lifecycle

Transport Layer 3-90

Chapter 3 outline
3.1 transport-layer 3.5 connection-oriented
services transport: TCP
3.2 multiplexing and • segment structure
demultiplexing • reliable data transfer
3.3 connectionless • flow control
transport: UDP • connection management
3.4 principles of reliable 3.6 principles of congestion
data transfer control
3.7 TCP congestion control

Transport Layer 3-91

Principles of congestion control
congestion:
 informally: “too many sources sending too much
data too fast for network to handle”
 different from flow control!
 manifestations:
• lost packets (buffer overflow at routers)
• long delays (queueing in router buffers)
 a top-10 problem!

Transport Layer 3-92

Causes/costs of congestion: scenario 1
original data: lin throughput: lout
 two senders, two
receivers Host A

 one router, infinite buffers unlimited shared

 output link capacity: R output link buffers

 no retransmission

Host B

R/2

delay
lout

lin R/2 lin R/2

 maximum per-connection  large delays as arrival rate, lin,
throughput: R/2 approaches capacity
Transport Layer 3-93
Causes/costs of congestion: scenario 2
 one router, finite buffers
 sender retransmission of timed-out packet
• application-layer input = application-layer output: lin = lout
• transport-layer input includes retransmissions : l‘ in lin

lin : original data

lout
l'in: original data, plus
retransmitted data

Host A

finite shared output

Host B
link buffers
Transport Layer 3-94
Causes/costs of congestion: scenario 2
R/2
idealization: perfect
knowledge

lout
 sender sends only when
router buffers available
lin R/2

lin : original data

lout
copy l'in: original data, plus
retransmitted data

A free buffer space!

finite shared output

Host B
link buffers
Transport Layer 3-95
Causes/costs of congestion: scenario 2
Idealization: known loss
packets can be lost,
dropped at router due
to full buffers
 sender only resends if
packet known to be lost

lin : original data

lout
copy l'in: original data, plus
retransmitted data

A
no buffer space!

Host B
Transport Layer 3-96
Causes/costs of congestion: scenario 2
Idealization: known loss R/2
packets can be lost,
dropped at router due when sending at R/2,
some packets are

lout
to full buffers retransmissions but

 sender only resends if

asymptotic goodput
is still R/2 (why?)
packet known to be lost lin R/2

lin : original data

lout
l'in: original data, plus
retransmitted data

A
free buffer space!

Host B
Transport Layer 3-97
Causes/costs of congestion: scenario 2
Realistic: duplicates R/2
 packets can be lost, dropped at
router due to full buffers when sending at R/2,
some packets are

lout
 sender times out prematurely, retransmissions

sending two copies, both of including duplicated

that are delivered!
which are delivered lin R/2

lin
timeout
copy l'in lout

A
free buffer space!

Host B
Transport Layer 3-98
Causes/costs of congestion: scenario 2
Realistic: duplicates R/2
 packets can be lost, dropped at
router due to full buffers when sending at R/2,
some packets are

lout
 sender times out prematurely, retransmissions

sending two copies, both of including duplicated

that are delivered!
which are delivered lin R/2

“costs” of congestion:
 more work (retrans) for given “goodput”
 unneeded retransmissions: link carries multiple copies of pkt
• decreasing goodput

Transport Layer 3-99

Causes/costs of congestion: scenario 3
 four senders Q: what happens as lin and lin’
increase ?
 multihop paths
A: as red lin’ increases, all arriving
 timeout/retransmit blue pkts at upper queue are
dropped, blue throughput g 0
Host A
lin : original data lout
Host B
l'in: original data, plus
retransmitted data
finite shared output
link buffers

Host D
Host C

Transport Layer 3-100

Causes/costs of congestion: scenario 3

C/2
lout

lin’ C/2

another “cost” of congestion:

 when packet dropped, any “upstream
transmission capacity used for that packet was
wasted!

Transport Layer 3-101

Chapter 3 outline
3.1 transport-layer 3.5 connection-oriented
services transport: TCP
3.2 multiplexing and • segment structure
demultiplexing • reliable data transfer
3.3 connectionless • flow control
transport: UDP • connection management
3.4 principles of reliable 3.6 principles of congestion
data transfer control
3.7 TCP congestion control

Transport Layer 3-102

Van (the Man) Jacobson

Transport Layer 3-103

 Congestion Collapse in the 1980s
 Early TCP used a fixed size sliding
window (e.g., 8 packets)
• Initially fine for reliability
 But something strange happened as
the ARPANET grew
• Links stayed busy but transfer rates
fell by orders of magnitude!

Computer Networks 104

 Congestion Collapse (2)
 Queues became full, retransmissions
clogged the network, and goodput fell

Congestion
collapse

Computer Networks 105

Transport Layer 3-106
 TCP Tahoe/Reno
 Avoid congestion collapse without changing routers
(or even receivers)
 Idea is to fix timeouts and introduce a congestion
window (cwnd) over the sliding window to limit
queues/loss
 TCP Tahoe/Reno implements AIMD by adapting
cwnd using packet loss as the network feedback
signal

Computer Networks 107

 TCP Tahoe/Reno (2)
 TCP behaviors we will study:
• ACK clocking
• Adaptive timeout (mean and
variance)
• Slow-start
• Fast Retransmission
• Fast Recovery

 Together, they implement

AIMD
Computer Networks 108
 TCP Timeline
TCP Reno
(Jacobson, ‘90)
TCP/IP “flag day”
3-way handshake (BSD Unix 4.2, ‘83)
(Tomlinson, ‘75)
TCP Tahoe
TCP and IP (Jacobson, ’88)
(RFC 791/793, ‘81)

Origins of “TCP”
(Cerf & Kahn, ’74)
Congestion collapse
Observed, ‘86
1988

1970 1975 1980 1985 1990

...
Pre-history Congestion control

Computer Networks 109

 TCP Timeline (2)
ECN Background TCP LEDBAT
Router support (Floyd, ‘94) (IETF ’08)
Delay TCP Vegas
based (Brakmo, ‘93)
Compound TCP
(Windows, ’07)
TCP with SACK
(Floyd, ‘96) FAST TCP
(Low et al., ’04)
TCP CUBIC
TCP Reno (Linux, ’06)
(Jacobson, ‘90) TCP New Reno
(Hoe, ‘95) TCP BIC
(Linux, ‘04

1990 1995 2000 2005 2010

... ...
Classic congestion control Diversification

Computer Networks 110

Nature of Congestion
 Routers/switches have internal buffering for
contention

Input ... Output

...
...

...
Fabric Output Buffer
Input Buffer

Computer Networks 111

Nature of Congestion (2)
 Simplified view of per port output queues
• Typically FIFO (First In First Out), discard when full

Router

Router

Queued
(FIFO) Queue Packets

Computer Networks 112

TCP congestion control:
 goal: TCP sender should transmit as fast as possible,
but without congesting network
 Q: how to find rate just below congestion level
 decentralized: each TCP sender sets its own rate, based
on implicit feedback:
 ACK: segment received (a good thing!), network not
congested, so increase sending rate
 lost segment: assume loss due to congested network,
so decrease sending rate

Transport Layer 3-114

TCP congestion control: bandwidth probing
 “probing for bandwidth”: increase transmission rate on
receipt of ACK, until eventually loss occurs, then
decrease transmission rate
 continue to increase on ACK, decrease on loss (since available
bandwidth is changing, depending on other connections in
network)
ACKs being received,
X loss, so decrease rate
so increase rate
X
X
X
sending rate

TCP’s
X “sawtooth”
behavior

time

 Q: how fast to increase/decrease?

 details to follow Transport Layer 3-115
TCP Congestion Control: details

 sender limits rate by limiting number of

unACKed bytes “in pipeline”:
LastByteSent-LastByteAcked  cwnd
• cwnd: differs from rwnd (how, why?)
• sender limited by min(cwnd,rwnd)
 roughly, cwnd
bytes

cwnd
rate = bytes/sec
RTT
RTT

 cwnd is dynamic, function of perceived ACK(s)

network congestion

Transport Layer 3-116

TCP Congestion Control: more details

Segment loss event: ACK received: increase rate

reduce rate
 slowstart phase:
 timeout: no response from
receiver  increase exponentially fast
• cut cwnd to 1 (despite name) at
connection start, or
 3 duplicate ACKs: at least following timeout
some segments getting
through (recall fast  congestion avoidance:
retransmit)  increase linearly
• cut cwnd in “half”, less
aggressively than on timeout

Transport Layer 3-117

 TCP Slow Start
 when connection begins, cwnd = 1
MSS
Host A Host B
• example: MSS = 500 bytes &
RTT = 200 msec
• initial rate = ?

RTT

time

Transport Layer 3-118

 TCP Slow Start
 when connection begins, cwnd = 1
MSS
Host A Host B
• example: MSS = 500 bytes &
RTT = 200 msec
• initial rate = 20 kbps

RTT
 available bandwidth may be >>
MSS/RTT
• desirable to quickly ramp up to
respectable rate
 increase rate exponentially until first
loss event or when threshold
reached
• double cwnd every RTT
• done by incrementing cwnd by
1 for every ACK received
time

Transport Layer 3-119

Transitioning into/out of slowstart
ssthresh: cwnd threshold maintained by TCP
 on loss event: set ssthresh to cwnd/2
• remember (half of) TCP rate when congestion last occurred
 when cwnd >= ssthresh: transition from slowstart to congestion avoidance
phase

duplicate ACK
dupACKcount++ new ACK
cwnd = cwnd+MSS
dupACKcount = 0
L transmit new segment(s),as allowed
cwnd = 1 MSS
ssthresh = 64 KB cwnd > ssthresh
dupACKcount = 0 slow L congestion
start timeout avoidance
ssthresh = cwnd/2
cwnd = 1 MSS
timeout dupACKcount = 0
retransmit missing segment
ssthresh = cwnd/2
cwnd = 1 MSS
dupACKcount = 0
retransmit missing segment

Transport Layer 3-120

 TCP: congestion avoidance
 when cwnd >
ssthresh grow cwnd
linearly
• approach possible
congestion slower than in
slowstart
• increase cwnd by
1 MSS per RTT
• implementation:
cwnd = cwnd + MSS●(MSS/cwnd)
for each ACK received
AIMD
 ACKs: increase cwnd
by 1 MSS per RTT:
additive increase
 loss: cut cwnd in half
(non-timeout-detected
loss ): multiplicative
decrease
AIMD: Additive Increase
Multiplicative Decrease
Transport Layer 3-121
 TCP congestion control: additive increase
multiplicative decrease
 approach: sender increases transmission rate (window
size), probing for usable bandwidth, until loss occurs
• additive increase: increase cwnd by 1 MSS every
RTT until loss detected
• multiplicative decrease: cut cwnd in half after loss
additively increase window size …
…. until loss occurs (then cut window in half)
congestion window size
cwnd: TCP sender

AIMD saw tooth

behavior: probing
for bandwidth

time
Transport Layer 3-122
TCP Congestion Control: details
sender sequence number space
cwnd TCP sending rate:
 roughly: send cwnd
bytes, wait RTT for
last byte last byte
ACKS, then send
ACKed sent, not-
yet ACKed
sent more bytes
(“in-
flight”) cwnd
 sender limits transmission: rate ~
~
RTT
bytes/sec

LastByteSent- < cwnd

LastByteAcked

 cwnd is dynamic, function

of perceived network
congestion
Transport Layer 3-123
TCP Slow Start
Host A Host B
 when connection begins,
increase rate
exponentially until first
loss event:

RTT
• initially cwnd = 1 MSS
• double cwnd every RTT
• done by incrementing
cwnd for every ACK
received
 summary: initial rate is
slow but ramps up
exponentially fast time

Transport Layer 3-124

 TCP: detecting, reacting to loss
 loss indicated by timeout:
• cwnd set to 1 MSS;
• window then grows exponentially (as in slow start)
to threshold, then grows linearly
 loss indicated by 3 duplicate ACKs: TCP RENO
• dup ACKs indicate network capable of delivering
some segments
• cwnd is cut in half window then grows linearly
 TCP Tahoe always sets cwnd to 1 (timeout or 3
duplicate acks)

Transport Layer 3-125

 TCP: switching from slow start to CA
Q: when should the
exponential
increase switch to
linear?
A: when cwnd gets
to 1/2 of its value
before timeout.

Implementation:
 variable ssthresh
 on loss event, ssthresh
is set to 1/2 of cwnd just
before loss event

* Check out the online interactive exercises for more

examples: http://gaia.cs.umass.edu/kurose_ross/interactive/ Transport Layer 3-126
TCP congestion control FSM: overview

slow cwnd > ssthresh

congestion
start avoidance
loss:
timeout
loss:
timeout

loss: new ACK loss:

timeout 3dupACK
fast
loss: recovery
3dupACK

Transport Layer 3-127

Summary: TCP Congestion Control
New
New ACK!
ACK! new ACK
duplicate ACK
dupACKcount++ new ACK
.
cwnd = cwnd + MSS (MSS/cwnd)
dupACKcount = 0
cwnd = cwnd+MSS transmit new segment(s), as allowed
dupACKcount = 0
L transmit new segment(s), as allowed
cwnd = 1 MSS
ssthresh = 64 KB cwnd > ssthresh
dupACKcount = 0 slow L congestion
start timeout avoidance
ssthresh = cwnd/2
cwnd = 1 MSS duplicate ACK
timeout dupACKcount = 0 dupACKcount++
ssthresh = cwnd/2 retransmit missing segment
cwnd = 1 MSS
dupACKcount = 0
retransmit missing segment
timeout
New
ACK!
ssthresh = cwnd/2
cwnd = 1 New ACK
dupACKcount = 0
cwnd = ssthresh dupACKcount == 3
dupACKcount == 3 retransmit missing segment dupACKcount = 0
ssthresh= cwnd/2 ssthresh= cwnd/2
cwnd = ssthresh + 3 cwnd = ssthresh + 3
retransmit missing segment retransmit missing segment
fast
recovery
duplicate ACK
cwnd = cwnd + MSS
transmit new segment(s), as allowed

Transport Layer 3-128

TCP throughput
 avg. TCP thruput as function of window size, RTT?
• ignore slow start, assume always data to send
 W: window size (measured in bytes) where loss occurs
• avg. window size (# in-flight bytes) is ¾ W
• avg. thruput is 3/4W per RTT
3 W
avg TCP thruput = bytes/sec
4 RTT

W/2

Transport Layer 3-129

TCP Futures: TCP over “long, fat pipes”

BANDWIDTH x DELAY

LFN (“elephant”)
 BD > 105 Bytes
How much data should be pushed in the pipe in order to never stall?

Transport Layer 3-130

TCP Futures: TCP over “long, fat pipes”

 example: 1500 byte segments, 100ms RTT, want

10 Gbps throughput
 requires W = 83,333 in-flight segments
 throughput in terms of segment loss probability, L
[Mathis 1997]:
1.22 . MSS
TCP throughput =
RTT L

➜ to achieve 10 Gbps throughput, need a loss rate of L

= 2·10-10 – a very small loss rate!
 new versions of TCP for high-speed

Transport Layer 3-131

TCP Fairness
fairness goal: if K TCP sessions share same
bottleneck link of bandwidth R, each should have
average rate of R/K

TCP connection 1

bottleneck
router
capacity R
TCP connection 2

Transport Layer 3-132

Why is TCP fair?
two competing sessions:
 additive increase gives slope of 1, as throughout increases
 multiplicative decrease decreases throughput proportionally

R equal bandwidth share

loss: decrease window by factor of 2

congestion avoidance: additive increase
loss: decrease window by factor of 2
congestion avoidance: additive increase

Connection 1 throughput R
Transport Layer 3-133
Fairness (more)
Fairness and UDP Fairness, parallel TCP
 multimedia apps often connections
do not use TCP  application can open
• do not want rate multiple parallel
throttled by congestion connections between
control
two hosts
 instead use UDP:
• send audio/video at
 web browsers do this
constant rate, tolerate  e.g., link of rate R with 9
packet loss existing connections:
• new app asks for 1 TCP, gets
rate R/10
• new app asks for 11 TCPs,
gets R/2

Transport Layer 3-134

Approaches towards congestion control
two broad approaches towards congestion control:
end-end congestion network-assisted
control: congestion control:
 no explicit feedback from  routers provide feedback to
network end systems
 congestion inferred from • single bit indicating
end-system observed loss, congestion (SNA, DECbit,
delay TCP/IP ECN, ATM)
 approach taken by TCP • explicit rate sender
should send at

Transport Layer 3-135

Explicit Congestion Notification (ECN)
network-assisted congestion control:
 two bits in IP header (ToS field) marked by network router
to indicate congestion
 congestion indication carried to receiving host
 receiver (seeing congestion indication in IP datagram) )
sets ECE bit on receiver-to-sender ACK segment to
notify sender of congestion
TCP ACK segment
source destination
application application
ECE=1
transport transport
network network
link link
physical physical

ECN=00 ECN=11

IP datagram
Transport Layer 3-136
Hall of fame

Transport Layer 3-137

Chapter 3: summary
 principles behind transport
layer services: next:
• multiplexing,  leaving the network
demultiplexing “edge” (application,
• reliable data transfer transport layers)
• flow control  into the network
• congestion control “core”
 instantiation,  two network layer
implementation in the chapters:
Internet • data plane
• UDP • control plane
• TCP
Transport Layer 3-138