Chapter 3

Transport Layer

Computer
Networking: A Top
Down Approach
6th edition
Jim Kurose, Keith Ross
Addison-Wesley
March 2012

Transport Layer 3-1

Chapter 3: Transport Layer
our goals:
v understand v learn about Internet
principles behind transport layer protocols:
transport layer § UDP: connectionless
services: transport
§ multiplexing, § TCP: connection-oriented
demultiplexing reliable transport
§ reliable data transfer § TCP congestion control
§ flow control
§ 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
v provide logical communication network
data link
between app processes physical

running on different hosts

lo
transport protocols run in

g
v

ica
end systems

le
nd
-e
§ send side: breaks app

nd
messages into segments,

tra
ns
passes to network layer

po
rt
§ rcv side: reassembles application
segments into messages, transport
network
passes to app layer data link
physical

v more than one transport

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

Transport Layer 3-5

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

delivery (TCP) data link

physical
network

§ congestion control network data link

lo
data link physical

g
physical

ica
§ flow control network

le
data link

nd
§ connection setup physical

-e
nd
network

unreliable, unordered data link

tra
v physical

ns
delivery: UDP

po
network
data link

rt
physical
§ no-frills extension of network
data link application
best-effort IP physical
network
data link
transport
network
data link
v 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 demultiplexing at receiver:
sockets, add transport header use header info to deliver
(later used for demultiplexing) 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
v 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
v host uses IP addresses & (payload)
port numbers to direct
segment to appropriate
TCP/UDP segment format
socket

Transport Layer 3-9

Connectionless demultiplexing
v recall: created socket has v recall: when creating
host-local port #: datagram to send into
DatagramSocket mySocket1 UDP socket, must specify
= new DatagramSocket(12534);
§ destination IP address
§ destination port #

v when host receives UDP IP datagrams with same

segment: dest. port #, but different
§ checks destination port # source IP addresses
in segment and/or source port
numbers will be directed
§ directs UDP segment to to same socket at dest
socket with that port #

Transport Layer 3-10

Connectionless demux: example
DatagramSocket
DatagramSocket serverSocket = new
DatagramSocket
mySocket2 = new DatagramSocket 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
v TCP socket identified v 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 v web servers have
v demux: receiver uses different sockets for
all 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]
v no frills, bare bones v UDP use:
Internet transport § streaming multimedia
protocol apps (loss tolerant, rate
v best effort service, sensitive)
UDP segments may be: § DNS
§ lost § SNMP
§ delivered out-of-order v reliable transfer over
to app
UDP:
v 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?
v no connection
application establishment (which can
data add delay)
(payload) v simple: no connection
state at sender, receiver
v small header size
UDP segment format v 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:
v treat segment contents, v compute checksum of
including header fields, received segment
as sequence of 16-bit v check if computed
integers
checksum equals checksum
v checksum: addition field value:
(one s complement
sum) of segment § NO - error detected
contents § YES - no error detected.
v sender puts checksum But maybe errors
value into UDP nonetheless? More later
checksum 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

Transport Layer 3-19

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 3-20

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

v characteristics of unreliable channel will determine

complexity of reliable data transfer protocol (rdt)
Transport Layer 3-21
Principles of reliable data transfer
v important in application, transport, link layers
§ top-10 list of important networking topics!

v characteristics of unreliable channel will determine

complexity of reliable data transfer protocol (rdt)
Transport Layer 3-22
Principles of reliable data transfer
v important in application, transport, link layers
§ top-10 list of important networking topics!

v characteristics of unreliable channel will determine

complexity of reliable data transfer protocol (rdt)
Transport Layer 3-23
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 3-24

rdt1.0: reliable transfer over a reliable channel
v underlying channel perfectly reliable
§ no bit errors
§ no loss of packets
v 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 3-25

rdt2.0: channel with bit errors
v underlying channel may flip bits in packet
§ checksum to detect bit errors
v 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
v new mechanisms in rdt2.0 (beyond rdt1.0):
§ error detection
during conversation?
§ receiver feedback: control msgs (ACK,NAK) rcvr-
>sender

Transport Layer 3-26

rdt2.0: channel with bit errors
v underlying channel may flip bits in packet
§ checksum to detect bit errors
v 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
v new mechanisms in rdt2.0 (beyond rdt1.0):
§ error detection
§ feedback: control msgs (ACK,NAK) from receiver to
sender

Transport Layer 3-27

 rdt2.0 in action
sender receiver sender receiver
send pkt0 pkt send pkt0 pkt
rcv pkt0 rcv pkt0
ack send ack ack send ack
rcv ack rcv ack
send pkt1 pkt send pkt1 pkt rcv pkt1 with
rcv pkt1 X
error bit error
ack send ack
rcv ack send nak
nak
send pkt2 pkt rcv nak
rcv pkt2 resend pkt1
ack send ack pkt
rcv pkt1
ack send ack
rcv ack
send pkt2 pkt
(a) no error rcv pkt2
ack send ack

(b) packet error

Transport Layer 3-28
rdt2.0 has a fatal flaw!
what happens if handling duplicates:
ACK/NAK corrupted? v sender retransmits
v sender doesn t know current pkt if ACK/NAK
what happened at corrupted
receiver!
v sender adds sequence
v can t just retransmit: number to each pkt
possible duplicate
v receiver discards (doesn t
deliver up) duplicate pkt
stop and wait
sender sends one packet,
then waits for receiver
response

Transport Layer 3-29

rdt2.1: discussion
sender: receiver:
v seq # added to pkt v must check if received

v two seq. # s (0,1) will packet is duplicate

suffice. Why? § state indicates whether
0 or 1 is expected pkt
v must check if received
seq #
ACK/NAK corrupted
v note: receiver can not
v 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 3-30

 rdt2.1 in action
sender receiver sender receiver
send pkt0 pkt 0 send pkt0 pkt 0
rcv pkt0 rcv pkt0
ack send ack ack send ack
rcv ack rcv ack
pkt 1 pkt 1
send pkt1 send pkt1
rcv pkt1 X discard pkt1
error error send nak
Xack send ack
X nak
discard ack/nak(?) discard ack/nak(?)
resend pkt1 resend pkt1
pkt 1 pkt 0
discard pkt1 rcv pkt1
send ack send ack
rcv ack ack rcv ack ack
send pkt2 pkt 0 send pkt2 pkt 1
rcv pkt0 rcv pkt0
send ack send ack
(a) ACK error (b) packet and NAK error
Transport Layer 3-31
rdt2.2: a NAK-free protocol
v same functionality as rdt2.1, using ACKs only
v instead of NAK, receiver sends ACK for last pkt
received OK
§ receiver must explicitly include seq # of pkt being ACKed
v duplicate ACK at sender results in same action as
NAK: retransmit current pkt

Transport Layer 3-32

 rdt2.2 in action
sender receiver sender receiver
send pkt0 pkt 0 send pkt0 pkt
rcv pkt0 rcv pkt0
ack 0 send ack ack 0 send ack
rcv ack rcv ack
pkt 1 pkt
send pkt1 send pkt1
rcv pkt1 X discard pkt1
error error send nak
Xack 1 send ack
X nak
discard ack/nak(?) discard ack/nak(?)
resend pkt1 resend pkt1
pkt 1 pkt 1
discard pkt1 rcv pkt1
send ack send ack
rcv ack ack 1 rcv ack ack
send pkt2 pkt 0 send pkt2 pkt 0
rcv pkt0 rcv pkt0
send ack send ack
(a) ACK error (b) packet and NAK error
Transport Layer 3-33
rdt3.0: channels with errors and loss

new assumption: approach: sender waits

underlying channel can reasonable amount of
also lose packets time for ACK
(data, ACKs) v retransmits if no ACK
§ checksum, seq. #, received in this time
ACKs, retransmissions v 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
v requires countdown timer

Transport Layer 3-34

 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 3-35
 rdt3.0 in action
sender receiver
sender receiver send pkt0 pkt0
send pkt0 pkt0 rcv pkt0
ack0 send ack0
rcv pkt0
send ack0 rcv ack0
ack0 send pkt1 pkt1
rcv ack0 rcv pkt1
send pkt1 pkt1
rcv pkt1 send ack1
ack1 ack1
send ack1
X
loss timeout
resend pkt1 pkt1
rcv pkt1
timeout
resend pkt1 pkt1 rcv ack1 pkt0 (detect duplicate)
rcv pkt1 send pkt0 send ack1
(detect duplicate) ack1
ack1 send ack1 rcv ack1 rcv pkt0
rcv ack1 ack0 send ack0
pkt0 send pkt0 pkt0
send pkt0 rcv pkt0
rcv pkt0 ack0 (detect duplicate)
ack0 send ack0 send ack0

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

Transport Layer 3-36

Performance of rdt3.0
v rdt3.0 is correct, but performance stinks
v 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
v network protocol limits use of physical resources!
Transport Layer 3-37
 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 3-38

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

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

selective repeat
Transport Layer 3-39
 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 3-40

Pipelined protocols: overview
Go-back-N: Selective Repeat:
v sender can have up to v sender can have up to N
N unacked packets in unack ed packets in
pipeline pipeline
v receiver only sends v rcvr sends individual ack
cumulative ack for each packet
§ doesn t ack packet if
there s a gap
v sender has timer for v 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 3-41

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

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

ACK
§ may receive duplicate ACKs (see receiver)
v timer for oldest in-flight pkt
v timeout(n): retransmit packet n and all higher seq # pkts in
window
Transport Layer 3-42
GBN: receiver extended FSM
ACK-only: always send ACK for correctly-received
pkt with highest in-order seq #
§ may generate duplicate ACKs
§ need only remember expectedseqnum
v out-of-order pkt:
§ discard (don t buffer): no receiver buffering!
§ re-ACK pkt with highest in-order seq #

Transport Layer 3-43

 GBN in action
sender window (N=4) sender receiver
012345678 send pkt0
012345678 send pkt1
send pkt2 receive pkt0, send ack0
012345678
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 3-44

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

Transport Layer 3-45

Selective repeat: sender, receiver windows

Transport Layer 3-46

 Selective repeat in action
sender window (N=4) sender receiver
012345678 send pkt0
012345678 send pkt1
send pkt2 receive pkt0, send ack0
012345678
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 3-47

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 3-48

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

v point-to-point: v full duplex data:

§ one sender, one receiver § bi-directional data flow
v reliable, in-order byte in same connection
steam: § MSS: maximum segment
size
§ no message
boundaries v connection-oriented:
v pipelined: § handshaking (exchange
of control msgs) inits
§ TCP congestion and sender, receiver state
flow control set window before data exchange
size
v flow controlled:
§ sender will not
overwhelm receiver
Transport Layer 3-49
 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 U A P R S F receive window
(generally not used) # bytes
checksum Urg data pointer rcvr willing
RST, SYN, FIN: options (variable length) to accept
connection estab
(setup, teardown
commands)
application
Internet data
checksum (variable length)
(as in UDP)

Transport Layer 3-50

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

first byte in segment s checksum

rwnd
urg pointer
data 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 3-51

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 3-52

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

Transport Layer 3-53

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

350

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

(milliseconds)

300

250
RTT (milliseconds)
RTT

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 3-54
SampleRTT Estimated RTT
TCP round trip time, timeout
v timeout interval: EstimatedRTT plus safety margin
§ large variation in EstimatedRTT -> larger safety margin
v estimate SampleRTT deviation from EstimatedRTT:
DevRTT = (1-b)*DevRTT +
b*|SampleRTT-EstimatedRTT|
(typically, b = 0.25)

TimeoutInterval = EstimatedRTT + 4*DevRTT

estimated RTT safety margin

Transport Layer 3-55

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 3-56

TCP reliable data transfer
v 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
v retransmissions § ignore flow control,
triggered by: congestion control
§ timeout events
§ duplicate acks

Transport Layer 3-57

TCP sender events:
data rcvd from app: timeout:
v create segment with v retransmit segment
seq # that caused timeout
v seq # is byte-stream v restart timer
number of first data ack rcvd:
byte in segment
v if ack acknowledges
v 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 3-58

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 3-59
TCP: retransmission scenarios
Host A Host B

Seq=92, 8 bytes of data

Seq=100, 20 bytes of data

ACK=100
timeout

X
ACK=120

Seq=120, 15 bytes of data

cumulative ACK
Transport Layer 3-60
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 3-61

TCP fast retransmit
v time-out period often
relatively long: TCP fast retransmit
§ long delay before if sender receives 3
resending lost packet ACKs for same data
v detect lost segments (( triple
triple duplicate
duplicate ACKs
ACKs ),),
via duplicate ACKs. resend unacked
§ sender often sends segment with smallest
many segments back- seq #
to-back
§ likely that unacked
§ if segment is lost, there segment lost, so don t
will likely be many wait for timeout
duplicate ACKs.

Transport Layer 3-62

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 3-63
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 3-64

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 3-65

TCP flow control
v 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
v sender limits amount of
unacked ( in-flight ) data to receiver-side buffering
receiver s rwnd value
v guarantees receive buffer
will not overflow
Transport Layer 3-66
 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 U A P R S F receive window
(generally not used) # bytes
checksum Urg data pointer rcvr willing
RST, SYN, FIN: options (variable length) to accept
connection estab
(setup, teardown
commands)
application
Internet data
checksum (variable length)
(as in UDP)

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

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

application application

connection state: ESTAB connection state: ESTAB

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

network network

Socket clientSocket = Socket connectionSocket =

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

Transport Layer 3-69

Agreeing to establish a connection

2-way handshake:
Q: will 2-way handshake
always work in
Let s talk
network?
ESTAB v variable delays
OK
ESTAB v retransmitted messages
(e.g. req_conn(x)) due to
message loss
v message reordering
choose x
req_conn(x)
v can t see other side
ESTAB
acc_conn(x)
ESTAB

Transport Layer 3-70

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 3-71
 TCP 3-way handshake

client state server state

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

Transport Layer 3-72

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

Transport Layer 3-73

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

Transport Layer 3-74

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 3-75

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

Transport Layer 3-76

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

v one router, infinite unlimited shared

buffers output link buffers

v output link capacity: R

v no retransmission
Host B

R/2

delay
lout

lin R/2 lin R/2

v maximum per-connection v large delays as arrival rate, lin,
throughput: R/2 approaches capacity
Transport Layer 3-77
Causes/costs of congestion: scenario 2
v one router, finite buffers
v sender retransmission of timed-out packet
§ application-layer input = application-layer output: lin =
lout
§ transport-layer input includes retransmissions : lin lin

lin : original data

l'in: original data, plus lout
retransmitted data

Host A

finite shared output

Host B
link buffers
Transport Layer 3-78
Causes/costs of congestion: scenario 2
R/2
idealization: perfect
knowledge

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

lin : original data

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

A free buffer space!

finite shared output

Host B
link buffers
Transport Layer 3-79
Causes/costs of congestion: scenario 2
Idealization: known loss
packets can be lost,
dropped at router due
to full buffers
v sender only resends if
packet known to be lost

lin : original data

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

A
no buffer space!

Host B
Transport Layer 3-80
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
asymptotic goodput
v sender only resends if 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 3-81
Causes/costs of congestion: scenario 2
Realistic: duplicates R/2
v packets can be lost, dropped
at router due to full buffers when sending at R/2,
some packets are

lout
v sender times out prematurely, retransmissions
including duplicated
sending two copies, both of that are delivered!
which are delivered lin R/2

lin
timeout
copy lout
l'in

A
free buffer space!

Host B
Transport Layer 3-82
Causes/costs of congestion: scenario 2
Realistic: duplicates R/2
v packets can be lost, dropped
at router due to full buffers when sending at R/2,
some packets are

lout
v sender times out prematurely, retransmissions
including duplicated
sending two copies, both of that are delivered!
which are delivered lin R/2

costs of congestion:
v more work (retrans) for given goodput
v unneeded retransmissions: link carries multiple copies of pkt
§ decreasing goodput

Transport Layer 3-83

Causes/costs of congestion: scenario 3
v four senders Q: what happens as lin and lin
increase ?
v multihop paths
A: as red lin increases, all arriving
v 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 3-84

Causes/costs of congestion: scenario 3

C/2
lout

lin C/2

another cost of congestion:

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

Transport Layer 3-85

Approaches towards congestion control

two broad approaches towards congestion control:

end-end congestion network-assisted

control: congestion control:
v no explicit feedback v routers provide
from network feedback to end systems
v congestion inferred § single bit indicating
from end-system congestion (SNA,
observed loss, delay DECbit, TCP/IP ECN,
v approach taken by ATM)
TCP § explicit rate for
sender to send at

Transport Layer 3-86

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 3-87

 TCP congestion control: additive increase
multiplicative decrease
v 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 3-88
TCP Congestion Control: details
sender sequence number space
cwnd TCP sending rate:
v 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
v sender limits transmission: rate ~
~
RTT
bytes/sec

LastByteSent- < cwnd

LastByteAcked

v cwnd is dynamic, function

of perceived network
congestion
Transport Layer 3-89
TCP Slow Start
Host A Host B
v when connection begins,
increase rate
exponentially until first one segm
ent
loss event:

RTT
§ initially cwnd = 1 MSS two segm
ents
§ double cwnd every RTT
§ done by incrementing
cwnd for every ACK four segm
ents
received
v summary: initial rate is
slow but ramps up
exponentially fast time

Transport Layer 3-90

 TCP: detecting, reacting to loss
v loss indicated by timeout:
§ cwnd set to 1 MSS;
§ window then grows exponentially (as in slow start)
to threshold, then grows linearly
v 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
v TCP Tahoe always sets cwnd to 1 (timeout or 3
duplicate acks)

Transport Layer 3-91

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

Transport Layer 3-92

Summary: TCP Congestion Control
New
New ACK!
ACK!
duplicate ACK
dupACKcount++ new ACK
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 3-93

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 3-94

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

R equal bandwidth share

Connection 2 throughput

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 3-95

Fairness (more)
Fairness and UDP Fairness, parallel TCP
v multimedia apps often connections
do not use TCP v application can open
§ do not want rate multiple parallel
throttled by congestion connections between two
control
hosts
v instead use UDP:
v web browsers do this
§ send audio/video at
constant rate, tolerate v 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 3-96

Chapter 3: summary
v principles behind
transport layer services:
§ multiplexing,
demultiplexing next:
v leaving the
§ reliable data transfer
network edge
§ flow control (application,
§ congestion control transport layers)
v instantiation, v into the network
implementation in the core
Internet
§ UDP
§ TCP
Transport Layer 3-97