Transport Layer: Computer Networking: A Top-Down Approach

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. 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!)
 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 material.
Computer Networking: A
For a revision history, see the slide note for this page.
Top-Down Approach
Thanks and enjoy! JFK/KWR 8th edition
Jim Kurose, Keith Ross
All material copyright 1996-2020
J.F Kurose and K.W. Ross, All Rights Reserved Pearson, 2020
Transport Layer: 3-1
Transport layer: overview
Our goal:
 understand principles  learn about Internet transport
behind transport layer layer protocols:
services: • UDP: connectionless transport
• multiplexing, • TCP: connection-oriented reliable
demultiplexing transport
• reliable data transfer • TCP congestion control
• flow control
• congestion control

Transport layer: roadmap
 Transport-layer services
 Multiplexing and demultiplexing
 Connectionless transport: UDP
 Principles of reliable data transfer
 Connection-oriented transport: TCP
 Principles of congestion control
 TCP congestion control
 Evolution of transport-layer
functionality
Transport services and protocols
application
transport
 provide logical communication mobile

network
network
data link
physical
between application processes national or global ISP
running on different hosts
log
ica
le
 transport protocols actions in end
nd
-e
systems:
nd
local or
tra
• sender: breaks application messages regional ISP
n sp
into segments, passes to network layer
ort
home network content
• receiver: reassembles segments into provider
network
messages, passes to application layer application
transport
datacenter
network
network
 two transport protocols available to data link

physical
Internet applications enterprise

network
• TCP, UDP
Transport vs. network layer services and protocols
household analogy:
12 kids in Ann’s house sending
letters to 12 kids in Bill’s house:
 hosts = houses
 processes = kids
 app messages = letters in
envelopes
 transport protocol = Ann and Bill
who demux to in-house siblings
 network-layer protocol = postal
service

Transport vs. network layer services and protocols
 network layer: logical household analogy:

communication between 12 kids in Ann’s house sending
hosts letters to 12 kids in Bill’s house:
 hosts = houses
 transport layer: logical  processes = kids
communication between  app messages = letters in
processes envelopes
• relies on, enhances, network  transport protocol = Ann and Bill
layer services who demux to in-house siblings
 network-layer protocol = postal
service

Transport Layer Actions
Sender:
application  is passed an application- app. msg
application
layer message
transport
 determines segment TTh htransport
app. msg
header fields values
network (IP)
 creates segment network (IP)
link
 passes segment to IP link
physical physical

Transport Layer Actions
Receiver:
application  receives segment from IP application
 checks header values
app. msg
transport  extracts application-layer transport
message
network (IP)  demultiplexes message up network (IP)
link to application via socket link
physical physical
Th app. msg

Two principal Internet transport protocols
application
transport
 TCP: Transmission Control Protocol mobile

network
network
data link
physical
national or global ISP
• reliable, in-order delivery
log
ica
le
• flow control
nd
-e
• connection setup
nd
local or
tra
regional ISP
 UDP: User Datagram Protocol
n sp
ort
• unreliable, unordered delivery home network content
provider
• no-frills extension of “best-effort” IP network
application
transport
datacenter
network
 services not available:

network
data link
physical
• delay guarantees enterprise

network
• bandwidth guarantees
Chapter 3: roadmap
functionality
HTTP server
client
application application
HTTP msg
transport
transport network transport

network link network
link physical link
physical physical

HTTP server
client
HTTP msg
Ht HTTP msg
transport

link physical link
physical physical

HTTP server
client
HTTP msg
Ht HTTP msg
transport
Hnnetwork
Ht HTTP msg
transport transport
link physical link
physical physical

HTTP server
client
transport

link physical link
physical physical
Hn Ht HTTP msg

HTTP server
client1 client2
application P-client1 P-client2 application
transport

link physical link
physical physical

Multiplexing/demultiplexing
multiplexing at sender: demultiplexing at receiver:
handle data from multiple use header info to deliver
sockets, add transport header received segments to correct
(later used for demultiplexing) socket
application
application P1 P2 application socket

P3 transport P4
process
link physical link
physical physical

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

Connectionless demultiplexing
Recall: when receiving host receives
 when creating socket, must UDP segment:
• checks destination port # in
specify host-local port #:
segment
DatagramSocket mySocket1 = new
DatagramSocket(12534); • directs UDP segment to socket
with that port #
 when creating datagram to
send into UDP socket, must
IP/UDP datagrams with same dest.
specify port #, but different source IP
• destination IP address addresses and/or source port
• destination port # numbers will be directed to same
socket at receiving host
Connectionless demultiplexing: an example
DatagramSocket
serverSocket = new
DatagramSocket
DatagramSocket mySocket2 = DatagramSocket mySocket1 =
new DatagramSocket (6428); new DatagramSocket (5775);
(9157); application
application P1 application
P3 P4
transport
transport transport
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: ?
Connection-oriented demultiplexing
 TCP socket identified by  server may support many
4-tuple: simultaneous TCP sockets:
• source IP address • each socket identified by its
• source port number own 4-tuple
• dest IP address • each socket associated with
• dest port number a different connecting client
 demux: receiver uses all
four values (4-tuple) to
direct segment to
appropriate socket
Connection-oriented demultiplexing: example
application
application P4 P5 P6 application
P1 P2 P3
transport
transport transport
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
Summary
 Multiplexing, demultiplexing: based on segment, datagram
header field values
 UDP: demultiplexing using destination port number (only)
 TCP: demultiplexing using 4-tuple: source and destination IP
addresses, and port numbers
 Multiplexing/demultiplexing happen at all layers

Chapter 3: roadmap
functionality
UDP: User Datagram Protocol
 “no frills,” “bare bones”
Why is there a UDP?
Internet transport protocol  no connection
establishment (which can
 “best effort” service, UDP add RTT delay)
segments may be:  simple: no connection state
• lost at sender, receiver
• delivered out-of-order to app  small header size
 connectionless:  no congestion control
 UDP can blast away as fast as
• no handshaking between UDP desired!
sender, receiver  can function in the face of
• each UDP segment handled congestion
independently of others
UDP: User Datagram Protocol
 UDP use:
 streaming multimedia apps (loss tolerant, rate sensitive)
 DNS
 SNMP
 HTTP/3
 if reliable transfer needed over UDP (e.g., HTTP/3):
 add needed reliability at application layer
 add congestion control at application layer

UDP: User Datagram Protocol [RFC 768]

UDP: Transport Layer Actions
SNMP client SNMP server
transport transport
(UDP) (UDP)
network (IP) network (IP)
link link
physical physical


UDP sender actions:
application  is passed an application- SNMP msg
application
layer message
transport  determines UDP segment UDPhtransport
UDP h SNMP msg
(UDP) header fields values (UDP)
network (IP)
 creates UDP segment network (IP)
link
 passes segment to IP link
physical physical


UDP receiver actions:
application  receives segment from IP application
 checks UDP checksum
transport
transport
SNMP msg header value
(UDP)  extracts application-layer (UDP)
h SNMP(IP)
network
UDP msg message network (IP)
 demultiplexes message up
link to application via socket link
physical physical

UDP segment header
32 bits
source port # dest port #
length checksum
application length, in bytes of

data UDP segment,
(payload) including header
data to/from
UDP segment format application layer

UDP checksum
Goal: detect errors (i.e., flipped bits) in transmitted segment
1st number 2nd number sum
Transmitted: 5 6 11
Received: 4 6 11
receiver-computed
checksum
= sender-computed
checksum (as received)

Internet checksum
Goal: detect errors (i.e., flipped bits) in transmitted segment
sender: receiver:
 treat contents of UDP  compute checksum of received
segment (including UDP header segment
fields and IP addresses) as
sequence of 16-bit integers  check if computed checksum equals
 checksum: addition (one’s checksum field value:
complement sum) of segment • not equal - error detected
content • equal - no error detected. But maybe
 checksum value put into errors nonetheless? More later ….
UDP checksum field
Internet checksum: an example
example: add two 16-bit integers
1110011001100110
1101010101010101
wraparound 11011101110111011
sum 1011101110111100
checksum 0100010001000011
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: h ttp://gaia.cs.umass.edu/kurose_ross/interactive/
Internet checksum: weak protection!
example: add two 16-bit integers
01
1110011001100110 10
1101010101010101
wraparound 11011101110111011 Even though
numbers have
sum 1011101110111100 changed (bit
flips), no change
checksum 0100010001000011 in checksum!

Summary: UDP
 “no frills” protocol:
• segments may be lost, delivered out of order
• best effort service: “send and hope for the best”
 UDP has its plusses:
• no setup/handshaking needed (no RTT incurred)
• can function when network service is compromised
• helps with reliability (checksum)
 build additional functionality on top of UDP in application layer
(e.g., HTTP/3)
Chapter 3: roadmap
functionality
Principles of reliable data transfer
sending receiving
process process
application data data
transport
reliable channel
reliable service abstraction

sending receiving sending receiving

process process process process
application data data application data data
transport transport
reliable channel
sender-side of receiver-side
reliable service abstraction reliable data of reliable data
transfer protocol transfer protocol
transport
network
unreliable channel
reliable service implementation

sending receiving
process process
transport
Complexity of reliable data reliable data
transfer protocol
of reliable data
transfer protocol
transfer protocol will depend
(strongly) on characteristics of transport
network
unreliable channel (lose, unreliable channel
corrupt, reorder data?)

sending receiving
process process
transport
reliable data of reliable data
Sender, receiver do not know transfer protocol transfer protocol
the “state” of each other, e.g.,
was a message received? transport
network
 unless communicated via a unreliable channel
message

Reliable data transfer protocol (rdt): interfaces
rdt_send(): called from above, deliver_data(): called by rdt to
(e.g., by app.). Passed data to deliver data to upper layer
deliver to receiver upper layer
sending receiving
process process
rdt_send() data data
deliver_data()
sender-side data receiver-side

implementation of implementation of
rdt reliable data packet rdt reliable data
transfer protocol transfer protocol
udt_send() Header data Header data rdt_rcv()
unreliable channel
udt_send(): called by rdt rdt_rcv(): called when packet
to transfer packet over Bi-directional communication over arrives on receiver side of
unreliable channel to receiver unreliable channel channel
Reliable data transfer: getting started
We will:
 incrementally develop sender, receiver sides of reliable data transfer
protocol (rdt)
 consider only unidirectional data transfer
• but control info will flow in 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 uniquely state state
determined by next 1 event
event 2
actions

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)

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

rdt2.0: channel with bit errors
 underlying channel may flip bits in packet
• checksum (e.g., Internet checksum) to detect bit errors
 the question: how to recover from errors?
How do humans recover from “errors” during conversation?

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
stop and wait

sender sends one packet, then waits for receiver response
rdt2.0: FSM specifications
rdt_send(data)
snkpkt = make_pkt(data, checksum)
udt_send(sndpkt)
rdt_rcv(rcvpkt) &&
Wait for Wait for isNAK(rcvpkt)
sender call from ACK or udt_send(sndpkt) rdt_rcv(rcvpkt) && corrupt(rcvpkt)
above NAK
udt_send(NAK)
rdt_rcv(rcvpkt) && isACK(rcvpkt)

Wait for
L
call from receiver
below
rdt_rcv(rcvpkt) && notcorrupt(rcvpkt)

extract(rcvpkt,data)
deliver_data(data)
udt_send(ACK)

rdt2.0: FSM specification
rdt_send(data)
udt_send(sndpkt)
rdt_rcv(rcvpkt) &&
above NAK
udt_send(NAK)

Wait for
L
call from receiver
below
Note: “state” of receiver (did the receiver get my

message correctly?) isn’t known to sender unless
somehow communicated from receiver to sender deliver_data(data)
 that’s why we need a protocol! udt_send(ACK)

rdt2.0: operation with no errors
rdt_send(data)
udt_send(sndpkt)
rdt_rcv(rcvpkt) &&
above NAK
udt_send(NAK)

Wait for
L call from receiver
below

deliver_data(data)
udt_send(ACK)

rdt2.0: corrupted packet scenario
rdt_send(data)
udt_send(sndpkt)
rdt_rcv(rcvpkt) &&
above NAK
udt_send(NAK)

Wait for
L call from receiver
below

deliver_data(data)
udt_send(ACK)

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

sender sends one packet, then
waits for receiver response
rdt2.1: sender, handling garbled ACK/NAKs
rdt_send(data)
sndpkt = make_pkt(0, data, checksum)
udt_send(sndpkt) rdt_rcv(rcvpkt) &&
(corrupt(rcvpkt) ||
Wait for Wait for isNAK(rcvpkt) )
call 0 from ACK or
NAK 0 udt_send(sndpkt)
above
rdt_rcv(rcvpkt)
rdt_rcv(rcvpkt)
&& notcorrupt(rcvpkt) &&
&& notcorrupt(rcvpkt)
isACK(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)

rdt2.1: receiver, handling garbled ACK/NAKs
&& has_seq0(rcvpkt)
deliver_data(data)
sndpkt = make_pkt(ACK, chksum)
udt_send(sndpkt)
rdt_rcv(rcvpkt) && rdt_rcv(rcvpkt) &&
(corrupt(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)
&& has_seq1(rcvpkt)
deliver_data(data)
sndpkt = make_pkt(ACK, chksum)
udt_send(sndpkt)

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

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
As we will see, TCP uses this approach to be NAK-free

rdt2.2: sender, receiver fragments
rdt_send(data)
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)
rdt_rcv(rcvpkt) && && isACK(rcvpkt,0)
(corrupt(rcvpkt) || L
has_seq1(rcvpkt)) Wait for receiver FSM
0 from
udt_send(sndpkt) below fragment
&& has_seq1(rcvpkt)
deliver_data(data)
sndpkt = make_pkt(ACK1, chksum)
udt_send(sndpkt) Transport Layer: 3-55
rdt3.0: channels with errors and loss
New channel assumption: underlying channel can also lose
packets (data, ACKs)
• checksum, sequence #s, ACKs, retransmissions will be of help …
but not quite enough
Q: How do humans handle lost sender-to-

receiver words in conversation?

rdt3.0: channels with errors and loss
Approach: sender waits “reasonable” amount of time for ACK
 retransmits if no ACK received in this time
 if pkt (or ACK) just delayed (not lost):
• retransmission will be duplicate, but seq #s already handles this!
• receiver must specify seq # of packet being ACKed
 use countdown timer to interrupt after “reasonable” amount of
time
timeout

rdt3.0 sender
rdt_send(data)
udt_send(sndpkt)
start_timer
Wait for Wait

call 0 from for
above ACK0
rdt_rcv(rcvpkt)
&& notcorrupt(rcvpkt) rdt_rcv(rcvpkt)
&& isACK(rcvpkt,1) && notcorrupt(rcvpkt)
stop_timer && isACK(rcvpkt,0)
stop_timer
Wait Wait for
for call 1 from
ACK1 above
rdt_send(data)
udt_send(sndpkt)
start_timer

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 0 from
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) || udt_send(sndpkt)
isACK(rcvpkt,0) ) start_timer
L

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
loss
ack1 send ack1
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

rdt3.0 in action
sender receiver
sender receiver send pkt0
pkt0
rcv pkt0
send pkt0 pkt0 send ack0
ack0
rcv pkt0 rcv ack0
ack0 send ack0 send pkt1 pkt1
rcv ack0 rcv pkt1
rcv pkt1 ack1
ack1 send ack1
X timeout
loss resend pkt1
pkt1 rcv pkt1
timeout
resend pkt1 pkt1
rcv pkt1 rcv ack1 (detect duplicate)
(detect duplicate)
ack1 send ack1 ack1 rcv pkt0
rcv ack1 rcv ack1 send ack0
send pkt0 pkt0 (ignore) ack0
rcv pkt0
ack0 send ack0 pkt1
(c) ACK loss (d) premature timeout/ delayed ACK

Performance of rdt3.0 (stop-and-wait)
 U sender: utilization – fraction of time sender busy sending
 example: 1 Gbps link, 15 ms prop. delay, 8000 bit packet

• time to transmit packet into channel:
L 8000 bits
Dtrans = R = = 8 microsecs
109 bits/sec

rdt3.0: stop-and-wait operation
sender receiver
first packet bit transmitted, t = 0
first packet bit arrives

RTT last packet bit arrives, send ACK
ACK arrives, send next

packet, t = RTT + L / R

rdt3.0: stop-and-wait operation
sender receiver
L/R L/R
Usender =
RTT + L / R
.008 RTT
=
30.008
= 0.00027
 rdt 3.0 protocol performance stinks!

 Protocol limits performance of underlying infrastructure (channel)

Quiz # 2 Date :20/10/2022
 Assume
• Protocol = stop and wait
• L=8000 bits
• R=1 Gbps (1G=109)
• Propagation delay=15 ms.
• ACK transmission time processing time at node  ignore
Note:
• Link Utilization = Data sent/Max. data that could be sent.
• Calculate how many bits we are sending per second (throughput)
• If we are paying RS. 10,000 for bandwidth, how much Extra is being paid (while
getting throughput less than bandwidth)

Quiz # 2 Date :20/10/2022
 Hint:
 300008 seconds=8000bits
 Throughput= per second
 Bandwidth= different

rdt3.0: pipelined protocols operation
pipelining: sender allows multiple, “in-flight”, yet-to-be-acknowledged
packets
• range of sequence numbers must be increased
• buffering at sender and/or receiver

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!

Go-Back-N: sender
 sender: “window” of up to N, consecutive transmitted but unACKed pkts
• k-bit seq # in pkt header
 cumulative ACK: ACK(n): ACKs all packets up to, including seq # n

• on receiving ACK(n): move window forward to begin at n+1
 timer for oldest in-flight packet
 timeout(n): retransmit packet n and all higher seq # packets in window
Go-Back-N: receiver
 ACK-only: always send ACK for correctly-received packet so far, with
highest in-order seq #
• may generate duplicate ACKs
• need only remember rcv_base
 on receipt of out-of-order packet:
• can discard (don’t buffer) or buffer: an implementation decision
• re-ACK pkt with highest in-order seq #
Receiver view of sequence number space:

received and ACKed
… … Out-of-order: received but not ACKed
rcv_base
Not received
Go-Back-N in action
sender window (N=4) sender receiver
012345678 send pkt0
012345678 send pkt1
send pkt2 receive pkt0, send ack0
012345678
send pkt3 Xloss receive pkt1, send ack1
012345678
(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

Selective repeat
 receiver individually acknowledges all correctly received packets
• buffers packets, as needed, for eventual in-order delivery to upper
layer
 sender times-out/retransmits individually for unACKed packets
• sender maintains timer for each unACKed pkt
 sender window
• N consecutive seq #s
• limits seq #s of sent, unACKed packets

Selective repeat: sender, receiver windows

Selective repeat: sender and receiver
sender receiver
data from above: packet n in [rcvbase, rcvbase+N-1]
 if next available seq # in  send ACK(n)
window, send packet  out-of-order: buffer
timeout(n):  in-order: deliver (also deliver
buffered, in-order packets),
 resend packet n, restart timer advance window to next not-yet-
ACK(n) in [sendbase,sendbase+N]: received packet
 mark packet n as received packet n in [rcvbase-N,rcvbase-1]
 ACK(n)
 if n smallest unACKed packet,
advance window base to next otherwise:
unACKed seq #  ignore

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
record ack3 arrived send ack4
pkt 2 timeout send ack5
012345678 send pkt2
012345678 (but not 3,4,5)
012345678 rcv pkt2; deliver pkt2,
012345678 pkt3, pkt4, pkt5; send ack2
Q: what happens when ack2 arrives?

Selective repeat:
sender window receiver window
(after receipt) (after receipt)
a dilemma!
0123012 pkt0
0123012 pkt1 0123012
0123012 pkt2 0123012
0123012
example: 0123012 pkt3
X
0123012
 seq #s: 0, 1, 2, 3 (base 4 counting) pkt0 will accept packet
with seq number 0
 window size=3 (a) no problem
0123012 pkt0
0123012 pkt1 0123012
0123012 pkt2 X 0123012
X 0123012
X
timeout
retransmit pkt0
0123012 pkt0
will accept packet
with seq number 0
(b) oops!
Selective repeat:
sender window receiver window
(after receipt) (after receipt)
a dilemma!
0123012 pkt0
0123012 pkt1 0123012
0123012 pkt2 0123012
0123012
example: 0123012 pkt3
X
 seq #s: 0, 1, 2, 3 (base 4 counting)  receiver can’t
0123012
pkt0 will accept packet
see sender side with seq number 0
 window size=3 (a) no problem
 receiver
behavior
identical in both
cases!
0something’s
123012 pkt0
0(very)
1 2 3 0 1wrong!
Q: what relationship is needed 2 pkt1
pkt2
0123012
X
between sequence # size and 0123012 0123012
X 0123012
window size to avoid problem timeout
X
in scenario (b)? retransmit pkt0
0123012 pkt0
will accept packet
with seq number 0
(b) oops!
Chapter 3: roadmap
• segment structure
• reliable data transfer
• flow control
• connection management
TCP: overview RFCs: 793,1122, 2018, 5681, 7323
 point-to-point:  cumulative ACKs
• one sender, one receiver  pipelining:
 reliable, in-order byte • TCP congestion and flow control
steam: set window size
• no “message boundaries"  connection-oriented:
 full duplex data: • handshaking (exchange of control
• bi-directional data flow in messages) initializes sender,
same connection receiver state before data exchange
• MSS: maximum segment size  flow controlled:
• sender will not overwhelm receiver

TCP segment structure
32 bits
source port # dest port # segment seq #: counting

ACK: seq # of next expected sequence number bytes of data into bytestream
byte; A bit: this is an ACK (not segments!)
acknowledgement number
length (of TCP header) head not
len used C EUAP R SF receive window flow control: # bytes
Internet checksum checksum Urg data pointer receiver willing to accept
options (variable
C, E: congestion notification length)
TCP options
application data sent by
RST, SYN, FIN: connection data application into
management (variable length) TCP socket

TCP sequence 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 data checksum

rwnd
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

ACKed yet ACKed but not usable
Q: how receiver handles out-of- (“in-flight”) yet sent
order segments outgoing segment from receiver
• A: TCP spec doesn’t say, - up

source port # dest port #
sequence number
to implementor acknowledgement number

A rwnd
checksum urg pointer
TCP sequence numbers, ACKs
Host A Host B
User types‘C’
Seq=42, ACK=79, data = ‘C’
host ACKs receipt of‘C’,
echoes back ‘C’
Seq=79, ACK=43, data = ‘C’
host ACKs receipt
of echoed ‘C’
Seq=43, ACK=80
simple telnet scenario

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

EstimatedRTT = (1- )*EstimatedRTT + *SampleRTT
 exponential weighted moving average (EWMA)
 influence of past sample decreases exponentially fast
RTT: gaia.cs.umass.edu to fantasia.eurecom.fr
 typical value:  = 0.125 350
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)
SampleRTT Estimated RTT
 timeout interval: EstimatedRTT plus “safety margin”
• large variation in EstimatedRTT: want a larger safety margin
TimeoutInterval = EstimatedRTT + 4*DevRTT
estimated RTT “safety margin”
 DevRTT: EWMA of SampleRTT deviation from EstimatedRTT:

DevRTT = (1-)*DevRTT + *|SampleRTT-EstimatedRTT|
(typically,  = 0.25)
TCP Sender (simplified)
event: data received from event: timeout
application  retransmit segment that
caused timeout
 create segment with seq #
 restart timer
 seq # is byte-stream number
of first data byte in segment
event: ACK received
 start timer if not already
running  if ACK acknowledges
• think of timer as for oldest
previously unACKed segments
unACKed segment • update what is known to be
ACKed
• expiration interval:
TimeOutInterval • start timer if there are still
unACKed segments
TCP Receiver: ACK generation [RFC 5681]
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

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 send cumulative
SendBase=120 ACK for 120
ACK=100
ACK=120
SendBase=120
lost ACK scenario premature timeout

TCP: retransmission scenarios
Host A Host B

ACK=100
X
ACK=120
cumulative ACK covers

for earlier lost ACK

TCP fast retransmit
Host A Host B
TCP fast retransmit
if sender receives 3 additional
Seq=92
ACKs for same data (“triple Seq=1
, 8 bytes
of data
duplicate ACKs”), resend unACKed 0 0, 20 b
ytes o
f data
segment with smallest seq # X
 likely that unACKed segment lost,
=100
so don’t wait for timeout ACK
=100
timeout
ACK
CK =100
A
=100
Receipt of three duplicate ACKs ACK
indicates 3 segments received Seq=100, 20 bytes of data
after a missing segment – lost

segment is likely. So retransmit!

Chapter 3: roadmap
• segment structure
• reliable data transfer
• flow control
• connection management
TCP flow control
application
Q: What happens if network Application removing
process
layer delivers data faster than data from TCP socket

buffers
application layer removes TCP socket
data from socket buffers? receiver buffers
TCP
code
Network layer
delivering IP datagram
payload into TCP
IP
socket buffers code
from sender
receiver protocol stack

TCP flow control
application
process

buffers
TCP
code
Network layer
delivering IP datagram
payload into TCP
IP
socket buffers code
from sender

TCP flow control
application
process

buffers
TCP
code
receive window
flow control: # bytes
receiver willing to accept IP
code
from sender

TCP flow control
application
process

buffers
TCP
flow control code
receiver controls sender, so

sender won’t overflow IP
code
receiver’s buffer by
transmitting too much, too fast
from sender

TCP flow control
 TCP receiver “advertises” free buffer
space in rwnd field in TCP header to application process
• RcvBuffer size set via socket
options (typical default is 4096 bytes) RcvBuffer buffered data
• many operating systems autoadjust
rwnd free buffer space
RcvBuffer
 sender limits amount of unACKed
(“in-flight”) data to received rwnd TCP segment payloads
 guarantees receive buffer will not TCP receiver-side buffering

overflow

TCP flow control
flow control: # bytes receiver willing to accept
 TCP receiver “advertises” free buffer

space in rwnd field in TCP header
• RcvBuffer size set via socket
receive window
options (typical default is 4096 bytes)
• many operating systems autoadjust
RcvBuffer
 sender limits amount of unACKed
(“in-flight”) data to received rwnd
 guarantees receive buffer will not
overflow
TCP segment format

Advertised Window Limits Rate
 Sender can send no faster than W/RTT bytes/sec
 Receiver only advertises more space when it has consumed

old arriving data
– Flow control: adjusting the sending rate to keep from

overwhelming a slow receiver
98
99
Starting and Ending a Connection:

TCP Handshakes
100
EstablishingAa TCPBConnection
SYN
CK
S YN A Each host tells its
ACK ISN to other host
Data
Data
 Three-way handshake to establish connection

• Host A sends a SYNchronize (open) to the host B
• Host B returns a SYN ACKnowledgment (SYN ACK)
• Host A sends an ACK to acknowledge the SYN ACK
101
TCP Header
Source port Destination port
Flags: SYN Sequence number

FIN Acknowledgment
RST
PSH HdrLen 0 Flags Advertised window
URG
ACK Checksum Urgent pointer
Options (variable)
Data
102
Step 1: A’s Initial SYN Packet
B’s port A’s port

Flags: SYN
A’s Initial Sequence Number
FIN
RST Acknowledgment
PSH
URG 20 0 Flags Advertised window
ACK
Checksum Urgent pointer
Options (variable)
A tells B it wants to open a connection…

103
Step 2: B’s SYN-ACK Packet

Flags: SYN
B’s Initial Sequence Number
FIN
RST A’s ISN plus 1
PSH
ACK
Options (variable)
B tells A it accepts, and is ready to hear the next byte…
… upon receiving this packet, A can start sending data

104
Step 3: A’s ACK of the SYN-ACK

Flags: SYN
Sequence number
FIN
RST B’s ISN plus 1
PSH
ACK
Options (variable)
A tells B it is okay to start sending…
… upon receiving this packet, B can start sending data

105
What if the SYN Packet Gets Lost?

 Suppose SYN packet gets lost
• Packet lost inside network, or
• Server rejects packet (e.g., listen queue is full)
 Eventually, no SYN-ACK arrives

• Sender sets timer and waits, retransmitting if needed
 How should TCP sender set timer?

• Sender has no estimate of RTT
• Some TCP stacks use default of 3 or 6 seconds
• Why reload button in your browser is useful:
Opens new connection and “retransmits” SYN in < 3-6 sec
106
Tearing
A
Down
B
the Connection
SY N • Closing a connection
CK
– Process done writing: invokes close()
SY N A
Setup – Once TCP sends all outstanding byte,
AC K
TCP sends a FINish message
Data
Data
• Receiving a FINish
A CK Data
– Process reading data from socket
Transfer
A CK – Eventually, read attempt returns EOF
FIN
• Tear-down is two-way
A CK
Teardown – FIN to close, but receive remaining
FIN
ACK – Other host ACKs the FIN
– Rest (RST) to close and not receive
remaining: error condition
Closing a TCP 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

An Simpler View of the Client Side
SYN (Send)
CLOSED
TIME_WAIT
SYN_SENT
Rcv. FIN,
Send ACK Rcv. SYN+ACK,
Send ACK
ESTABLISHED
FIN_WAIT2
Rcv. ACK, FIN_WAIT1 Send FIN

Send Nothing
Chapter 3: roadmap
functionality
Principles of congestion control
Congestion:
 informally: “too many sources sending too much data too fast for
network to handle”
 manifestations:
• long delays (queueing in router buffers)
• packet loss (buffer overflow at routers)
 different from flow control! congestion control:
 a top-10 problem! too many senders,
sending too fast
flow control: one sender

too fast for one receiver
Causes/costs of congestion: scenario 1
original data: lin throughput: lout
Simplest scenario:
Host A
 one router, infinite buffers
 input, output link capacity: R infinite shared
output link buffers
 two flows
R R
 no retransmissions needed
Host B
R/2
Q: What happens as
lout
delay
arrival rate lin
throughput:
approaches R/2?
lin R/2 lin R/2
maximum per-connection large delays as arrival rate
throughput: R/2 lin approaches capacity
 one router, finite buffers
 sender retransmits lost, timed-out packet
• application-layer input = application-layer output: lin = lout
• transport-layer input includes retransmissions : l’in lin
Host A lin : original data

l'in: original data, plus lout
retransmitted data
R R
Host B finite shared output

link buffers
Idealization: perfect knowledge R/2
 sender sends only when router buffers available
lout
throughput:
Host A lin : original data lin
copy l'in: original data, plus lout R/2
retransmitted data
free buffer space!
R R

link buffers
Idealization: some perfect knowledge
 packets can be lost (dropped at router) due to
full buffers
 sender knows when packet has been dropped:
only resends if packet known to be lost

copy l'in: original data, plus
retransmitted data
no buffer space!
R R

link buffers
Idealization: some perfect knowledge R/2
“wasted” capacity due
lout
 packets can be lost (dropped at router) due to to retransmissions
full buffers
throughput:
when sending at
 sender knows when packet has been dropped: R/2, some packets
only resends if packet known to be lost are needed
retransmissions
Host A lin : original data lin R/2

l'in: original data, plus
retransmitted data
free buffer space!
R R

link buffers
Realistic scenario: un-needed duplicates R/2
 packets can be lost, dropped at router due to
lout
full buffers – requiring retransmissions to un-needed
retransmissions
 but sender times can time out prematurely,
throughput:
sending two copies, both of which are delivered when sending at
R/2, some packets
are retransmissions,
including needed
and un-needed
Host A lin : original data lin
copy R/2 duplicates, that are
timeout
l'in: original data, plus delivered!
retransmitted data
free buffer space!
R R

link buffers
Realistic scenario: un-needed duplicates R/2
 packets can be lost, dropped at router due to
lout
full buffers – requiring retransmissions to un-needed
retransmissions
 but sender times can time out prematurely,
throughput:
sending two copies, both of which are delivered when sending at
R/2, some packets
are retransmissions,
including needed
and un-needed
lin R/2 duplicates, that are
delivered!
“costs” of congestion:
 more work (retransmission) for given receiver throughput
 unneeded retransmissions: link carries multiple copies of a packet
• decreasing maximum achievable throughput

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

R/2
lout
lin’ R/2
another “cost” of congestion:

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

Causes/costs of congestion: insights
 throughput can never exceed capacity
 delay increases as capacity approached
 loss/retransmission decreases effective

throughput
 un-needed duplicates further decreases

effective throughput
 upstream transmission capacity / buffering
wasted for packets lost downstream
Approaches towards congestion control
End-end congestion control:

 no explicit feedback from
network
 congestion inferred from ACKs
data data
ACKs
observed loss, delay
 approach taken by TCP

Network-assisted congestion
control: explicit congestion info
 routers provide direct feedback
to sending/receiving hosts with data data
ACKs
flows passing through congested ACKs
router
 may indicate congestion level or
explicitly set sending rate
 TCP ECN, ATM, DECbit protocols

Chapter 3: roadmap
functionality
TCP Congestion control
TCP reacts to congestion by reducing the sender window size.
Sender windows size determined by
Receiver window
Congestion window

Receiver window size is an advertisement of-
“How much data (in bytes) the receiver can receive without acknowledgement?”
1. Receiver Window Size-

Sender should not send data greater than receiver window size.
•Otherwise, it leads to dropping the TCP segments which causes TCP Retransmission.
•So, sender should always send data less than or equal to receiver window size.
•Receiver dictates its window size to the sender through TCP Header.

2. Congestion Window-

•Sender should not send data greater than congestion window size.
•Otherwise, it leads to dropping the TCP segments which causes TCP Retransmission.
•So, sender should always send data less than or equal to congestion window size.
•Different variants of TCP use different approaches to calculate the size of congestion window.
•Congestion window is known only to the sender and is not sent over the links.

Sender window size = Minimum (Receiver window size, Congestion window size)

TCP Congestion policy

 1. Slow Start Phase-

• Initially, sender sets congestion window size =
Maximum Segment Size (1 MSS).
• After receiving each acknowledgment, sender increases
the congestion window size by 1 MSS.
• In this phase, the size of congestion window increases
exponentially.
This phase continues until the congestion window size reaches

the slow start threshold.
Threshold
= Maximum number of TCP segments that receiver window can accommodate / 2
= (Receiver window size / Maximum Segment Size) / 2

2. Congestion Avoidance Phase-
 After reaching the threshold,

• Sender increases the congestion
window size linearly to avoid the
congestion.
• On receiving each acknowledgement,
sender increments the congestion
window size by 1.
until the congestion window size becomes equal to the receiver

window size

3. Congestion Detection Phase-
 When sender detects the loss of segments, it reacts in different ways depending on how the loss is detected-

 Case-01: Detection On Time Out-

• Time Out Timer expires before receiving the acknowledgement for a segment.
• This case suggests the stronger possibility of congestion in the network.
• There are chances that a segment has been dropped in the network.

 Reaction-

 In this case, sender reacts by-
• Setting the slow start threshold to half of the current congestion window size.
• Decreasing the congestion window size to 1 MSS. (sending rate Cut to 1 MSS (maximum segment size) when
loss detected by timeout (TCP Tahoe)
• Resuming the slow start phase.

3. Congestion Detection Phase-
 When sender detects the loss of segments, it reacts in different ways depending on how the loss is detected-


 Case-02: Detection On Receiving 3 Duplicate Acknowledgements-

• Sender receives 3 duplicate acknowledgements for a segment.
• This case suggests the weaker possibility of congestion in the network.
• There are chances that a segment has been dropped but few segments sent later may have reached.

 Reaction-

 In this case, sender reacts by-
• Setting the slow start threshold to half of the current congestion window size. (sending rate cut in half
(Multiplicative decrease) and linear increase/Additive increase)
• Decreasing the congestion window size to slow start threshold.
• Resuming the congestion avoidance phase.
TCP congestion control: AIMD
 approach: senders can increase sending rate until packet loss
(congestion) occurs, then decrease sending rate on loss event
Additive Increase Multiplicative Decrease
increase sending rate by 1 cut sending rate in half at
maximum segment size every each loss event
RTT until loss detected
TCP sender Sending rate
AIMD sawtooth
behavior: probing
for bandwidth
time Transport Layer: 3-132

TCP AIMD: more
Multiplicative decrease detail: sending rate is
 Cut in half on loss detected by triple duplicate ACK (TCP Reno)
 Cut to 1 MSS (maximum segment size) when loss detected by
timeout (TCP Tahoe)
Why AIMD?
 AIMD – a distributed, asynchronous algorithm – has been
shown to:
• optimize congested flow rates network wide!
• have desirable stability properties

TCP congestion control: details
sender sequence number space
cwnd TCP sending behavior:
 roughly: send cwnd bytes,
wait RTT for ACKS, then
send more bytes
last byte
available but cwnd
ACKed sent, but not-
not used
TCP rate ~
~ bytes/sec
yet ACKed RTT
(“in-flight”) last byte sent
 TCP sender limits transmission: LastByteSent- LastByteAcked < cwnd

 cwnd is dynamically adjusted in response to observed network
congestion (implementing TCP congestion control)
TCP slow start
Host A Host B
 when connection begins,
increase rate exponentially
one segm
until first loss event: ent
RTT
• initially cwnd = 1 MSS two segm
ents
• double cwnd every RTT
• done by incrementing cwnd
for every ACK received four segm
ents
 summary: initial rate is

slow, but ramps up
exponentially fast time

TCP: from slow start to congestion avoidance
Q: when should the exponential
increase switch to linear?
X
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
Summary: TCP congestion control
New
New ACK!
.
ACK! new ACK
duplicate ACK
cwnd = cwnd + MSS (MSS/cwnd)
dupACKcount++ new ACK 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
fast retransmit missing segment
recovery
duplicate ACK
cwnd = cwnd + MSS
transmit new segment(s), as allowed

TCP CUBIC
 Is there a better way than AIMD to “probe” for usable bandwidth?
 Insight/intuition:
• Wmax: sending rate at which congestion loss was detected
• congestion state of bottleneck link probably (?) hasn’t changed much
• after cutting rate/window in half on loss, initially ramp to to Wmax faster, but then
approach Wmax more slowly
Wmax classic TCP
TCP CUBIC - higher

Wmax/2 throughput in this
example

TCP CUBIC
 K: point in time when TCP window size will reach Wmax
• K itself is tuneable
 increase W as a function of the cube of the distance between current
time and K
• larger increases when further away from K
• smaller increases (cautious) when nearer K
 TCP CUBIC default Wmax
in Linux, most
TCP Reno
popular TCP for TCP CUBIC
popular Web TCP
sending
servers rate
time
t0 t1 t2 t3 t4
TCP and the congested “bottleneck link”
 TCP (classic, CUBIC) increase TCP’s sending rate until packet loss occurs
at some router’s output: the bottleneck link
source destination
TCP TCP
network network
link link
physical physical
packet queue almost
never empty, sometimes
overflows packet (loss)
bottleneck link (almost always busy)

TCP and the congested “bottleneck link”
 TCP (classic, CUBIC) increase TCP’s sending rate until packet loss occurs
at some router’s output: the bottleneck link
 understanding congestion: useful to focus on congested bottleneck link
insight: increasing TCP sending rate will

source not increase end-end throughout destination
with congested bottleneck
TCP TCP
network network
link link
physical physical
insight: increasing TCP

sending rate will
increase measured RTT
Goal: “keep the end-end pipe just full, but not fuller”
RTT
Delay-based TCP congestion control
Keeping sender-to-receiver pipe “just full enough, but no fuller”: keep
bottleneck link busy transmitting, but avoid high delays/buffering
# bytes sent in
measured last RTT interval
RTTmeasured throughput =
RTTmeasured
Delay-based approach:
 RTTmin - minimum observed RTT (uncongested path)
 uncongested throughput with congestion window cwnd is cwnd/RTTmin
if measured throughput “very close” to uncongested throughput
increase cwnd linearly /* since path not congested */
else if measured throughput “far below” uncongested throughout
decrease cwnd linearly /* since path is congested */
Delay-based TCP congestion control
 congestion control without inducing/forcing loss
 maximizing throughout (“keeping the just pipe full… ”) while keeping
delay low (“…but not fuller”)
 a number of deployed TCPs take a delay-based approach
 BBR deployed on Google’s (internal) backbone network

Explicit congestion notification (ECN)
TCP deployments often implement network-assisted congestion control:
 two bits in IP header (ToS field) marked by network router to indicate congestion
• policy to determine marking chosen by network operator
 congestion indication carried to destination
 destination sets ECE bit on ACK segment to notify sender of congestion
 involves both IP (IP header ECN bit marking) and TCP (TCP header C,E bit marking)
source TCP ACK segment
destination
ECE=1
TCP TCP
network network
link link
physical physical
ECN=10 ECN=11
IP datagram
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
TCP connection 2 router
capacity R

Q: is TCP Fair?
Example: two competing TCP sessions:
 additive increase gives slope of 1, as throughout increases
 multiplicative decrease decreases throughput proportionally
R equal bandwidth share

Is TCP fair?
Connection 2 throughput
A: Yes, under idealized

loss: decrease window by factor of 2 assumptions:
congestion avoidance: additive increase  same RTT
loss: decrease window by factor of 2
congestion avoidance: additive increase
 fixed number of sessions
only in congestion
avoidance
Connection 1 throughput R
Fairness: must all network apps be “fair”?
Fairness and UDP Fairness, parallel TCP
 multimedia apps often do not connections
use TCP  application can open multiple
• do not want rate throttled by parallel connections between two
congestion control hosts
 instead use UDP:  web browsers do this , e.g., link of
• send audio/video at constant rate, rate R with 9 existing connections:
tolerate packet loss
• new app asks for 1 TCP, gets rate R/10
 there is no “Internet police” • new app asks for 11 TCPs, gets R/2
policing use of congestion
control

Transport layer: roadmap
functionality
Evolving transport-layer functionality
 TCP, UDP: principal transport protocols for 40 years
 different “flavors” of TCP developed, for specific scenarios:
Scenario Challenges
Long, fat pipes (large data Many packets “in flight”; loss shuts down
transfers) pipeline
Wireless networks Loss due to noisy wireless links, mobility;
TCP treat this as congestion loss
Long-delay links Extremely long RTTs
Data center networks Latency sensitive
Background traffic flows Low priority, “background” TCP flows
 moving transport–layer functions to application layer, on top of UDP

• HTTP/3: QUIC
QUIC: Quick UDP Internet Connections
 application-layer protocol, on top of UDP
• increase performance of HTTP
• deployed on many Google servers, apps (Chrome, mobile YouTube app)
HTTP/2 HTTP/2 (slimmed)

Application HTTP/3
TLS QUIC
Transport TCP UDP
Network IP IP
HTTP/2 over TCP HTTP/2 over QUIC over UDP

QUIC: Quick UDP Internet Connections
adopts approaches we’ve studied in this chapter for
connection establishment, error control, congestion control
• error and congestion control: “Readers familiar with TCP’s loss
detection and congestion control will find algorithms here that parallel
well-known TCP ones.” [from QUIC specification]
• connection establishment: reliability, congestion control,
authentication, encryption, state established in one RTT
 multiple application-level “streams” multiplexed over single QUIC

connection
• separate reliable data transfer, security
• common congestion control
QUIC: Connection establishment
TCP handshake
(transport layer) QUIC handshake
data
TLS handshake
(security)
data
TCP (reliability, congestion control QUIC: reliability, congestion control,

state) + TLS (authentication, crypto authentication, crypto state
state)
 1 handshake
 2 serial handshakes

QUIC: streams: parallelism, no HOL blocking
HTTP HTTP
GET GET HTTP
application
GET
HTTP HTTP
GET GET
HTTP
GET QUIC QUIC QUIC QUIC QUIC QUIC
encrypt encrypt encrypt encrypt encrypt encrypt
QUIC QUIC QUIC QUIC QUIC QUIC
TLS encryption TLS encryption RDT RDT RDT RDT
error!
RDT RDT
QUIC Cong. Cont. QUIC Cong. Cont.

transport
TCP RDT TCP

error! RDT
TCP Cong. Contr. TCP Cong. Contr. UDP UDP
(a) HTTP 1.1 (b) HTTP/2 with QUIC: no HOL blocking

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

Additional Chapter 3 slides

Go-Back-N: 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
Go-Back-N: receiver extended FSM
any other event
udt_send(sndpkt) rdt_rcv(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 packet with highest

in-order seq #
• may generate duplicate ACKs
• need only remember expectedseqnum
 out-of-order packet:
• discard (don’t buffer): no receiver buffering!
• re-ACK pkt with highest in-order seq # Transport Layer: 3-157
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
}
TCP 3-way handshake FSM
closed
Socket connectionSocket =
welcomeSocket.accept();
L Socket clientSocket =
newSocket("hostname","port number");
SYN(x)
SYNACK(seq=y,ACKnum=x+1) SYN(seq=x)
create new socket for communication
back to client
listen
SYN
SYN sent
rcvd
SYNACK(seq=y,ACKnum=x+1)
ESTAB
ACK(ACKnum=y+1) ACK(ACKnum=y+1)
L

Closing a TCP 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

TCP throughput
 avg. TCP thruput as function of window size, RTT?
• ignore slow start, assume there is 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
W/2
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!
 versions of TCP for long, high-speed scenarios

Transport Layer: Computer Networking: A Top-Down Approach

Uploaded by

Document Informationclick to expand document information

Document Informationclick to expand document information

Copyright:

Available Formats

Transport Layer: Computer Networking: A Top-Down Approach

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Transport Layer: Computer Networking: A Top-Down Approach

Uploaded by

Copyright:

Available Formats

Chapter 3

Transport Layer: 3-2

 provide logical communication mobile

between application processes national or global ISP

running on different hosts

 two transport protocols available to data link

Internet applications enterprise

Transport Layer: 3-5

 network layer: logical household analogy:

Transport Layer: 3-6

Transport Layer: 3-7

link to application via socket link

Transport Layer: 3-8

 TCP: Transmission Control Protocol mobile

 services not available:

• delay guarantees enterprise

transport network transport

Transport Layer: 3-11

transport network transport

Transport Layer: 3-12

Transport Layer: 3-13

transport network transport

Transport Layer: 3-14

transport network transport

Transport Layer: 3-15

application P1 P2 application socket

Transport Layer: 3-16

Transport Layer: 3-17

source port: 6428 source port: ?

source port: 9157 source port: ?

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

Transport Layer: 3-22

Transport Layer: 3-25

Transport Layer: 3-26

SNMP client SNMP server

network (IP) network (IP)

Transport Layer: 3-27

SNMP client SNMP server

(UDP) header fields values (UDP)

Transport Layer: 3-28

SNMP client SNMP server

Transport Layer: 3-29

application length, in bytes of

Transport Layer: 3-30

Transport Layer: 3-31

Transport Layer: 3-34

reliable service abstraction

Transport Layer: 3-37

sending receiving sending receiving

reliable service implementation

Transport Layer: 3-38

Transport Layer: 3-39

Transport Layer: 3-40

sender-side data receiver-side

udt_send() Header data Header data rdt_rcv()

Transport Layer: 3-42

 separate FSMs for sender, receiver: