transport layer advanced features1 chapter 2 transport layer advance features some of these slides...

Transport Layer Advanced Features 1

Chapter 2Transport Layer Advance Features

Some of these slides are copyrighted by:

Computer Networking: A Top Down Approach 5th edition. Jim Kurose, Keith RossAddison-Wesley, April 2009.

Departamento deTecnología Electrónica


Chapter 2 outline

2.1 Transport-layer services

2.2 Connection-oriented transport: TCP

2.3 Principles of congestion control

2.4 TCP congestion control

2.5 Stream Control Transmission Protocol


Chapter 2 outline







Transport services and protocols

provide logical communication between app processes running on different hosts

transport protocols run in end systems breaks app messages into segments, passes to

network layer

NetworkHost A Host B


Internet transport-layer protocols more than one transport protocol available to apps

Internet: TCP and UDP

FTP HTTP SMTP DNS TFTP

TCP UDP

IP

Link layer (LLC y MAC)

Physical Layer

Application

Transport

Network


Internet transport-layer protocols more than one

transport protocol available to apps Internet: TCP

and UDP

TCP UDP

Connection- oriented

Non-connectionoriented

Reliable Unreliable

Segment grouping Unfragmented messages

Rcv orders segments

User datagram

ACKs and timers No ACKs

Flow control No flow control

Congestion control No congestion control


Internet transport-layer protocols Port: identifies application Port numbers:

http://www.iana.org/assignments/port-numbers

Application protocol

Port numbers Transport protocol

FTP 20, 21 TCP

Telnet 23 TCP

SMTP 25 TCP

DNS 53 UDP (TCP (*))

TFTP 69 UDP

HTTP 80 TCP

POP3 110 TCP

RIP 520 UDP


Chapter 2 outline







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

full duplex data: bi-directional data flow

in same connection MSS: maximum

segment size

connection-oriented: handshaking (exchange

of control msgs) init’s sender, receiver state before data exchange

flow controlled: sender will not

overwhelm receiver

point-to-point: one sender, one

receiver

reliable, in-order byte stream: no “message

boundaries”

pipelined: TCP congestion and flow

control set window size

send & receive bufferssocketdoor

T C Psend buffer

T C Preceive buffer

socketdoor

segm ent

applicationwrites data

applicationreads data


TCP segment structure

source port # dest port #

32 bits

applicationdata

(variable length)

sequence number

acknowledgement numberReceive window

Urg data pnterchecksum

FSRPAUheadlen

notused

Options (variable length)

URG: urgent data (generally not used)

ACK: ACK #valid

PSH: push data now(generally not used)

RST, SYN, FIN:connection estab(setup, teardown

commands)

# bytes rcvr willingto accept

countingby bytes of data(not segments!)

Internetchecksum

(as in UDP)


TCP seq. #’s and ACKsHost A Host B

Seq=M, SYN=1

Seq=N, ACK=M+1, SYN=1

Seq=M+1, ACK=N+1

Client starts active open

Client confirms

server open

Server is in passive open, starts connection

and confirms client open

timeConnection established


Chapter 2 outline







Principles of Congestion Control

Congestion: informally: “too many sources sending too

much data too fast for network to handle” different from flow control! manifestations:

lost packets (buffer overflow at routers) long delays (queueing in router buffers)


Causes/costs of congestion

Two tx: IN = Sending rate

OUT = Receiving rate

C= Router capability

Queue theory

unlimited shared output link buffers

Host Ain : original data

Host B

out

large delays when congested

maximum achievable throughput


Supposition: Tx retx lost packets Now, IN is not tx -> higher rate IN’ Congestion costs:

more work (retx) unneeded retransmissions: link carries multiple copies of pkt

Causes/costs of congestion:

unlimited shared output link buffers


Host B

out


Causes/costs of congestion Two tx and the same dest Supposition: congestion in the router between A and dest Another cost of congestion:

when a packet is dropped, any upstream transmission capacity used for that packet was wasted

finite shared output link buffers


Host B

out

'in : original data, plus retransmitted data


Causes/costs of congestion Congestion costs: a summary

Big delays when the incoming rate to a router is near to the link capability.

The tx must retx the lost segments due to the router overflow.

Unnecessary BW is used for the retx of duplicated copies of segments (due to delays).

Upstream transmission capacity used for a packet is wasted when it is discarded.


Approaches towards congestion control

End-end congestion control:

no explicit feedback from network

congestion inferred from end-system observed loss, delay

approach taken by TCP

Network-assisted congestion control:

routers provide feedback to end systems E.g. single bit

indicating congestion

Two broad approaches towards congestion control:


Chapter 2 outline







TCP Congestion Control

sender limits transmission:

LastByteSent-LastByteAcked CongWin

How does sender perceive congestion? loss event = timeout

or 3 duplicate acks TCP sender reduces

rate (CongWin) after loss event

Host A

tim

eout

Host B

time

X

resend 2nd segment

Host A

Seq=92, 8 bytes data

ACK=100

loss

tim

eout

lost ACK scenario

Host B

XSeq=92, 8 bytes data

ACK=

100

timeResending a

segment after triple duplicate ACK


TCP congestion control: additive increase, multiplicative decrease (AIMD)

8 Kbytes

16 Kbytes

24 Kbytes

time

congestionwindow

Approach: increase transmission rate (window size), probing for usable bandwidth, until loss occurs additive increase: increase CongWin by 1

MSS every RTT until loss detected multiplicative decrease: cut CongWin in half

after loss

timecong

estio

n w

indo

w s

ize

Saw toothbehavior: probing

for bandwidth


TCP Congestion Control: details

Roughly,

When connection begins, CongWin = 1 MSS Example: MSS = 500 bytes & RTT

= 200 msec initial rate = 20 kbps

three phases: slow start (SS) Congestion avoidance

(CA): e.g. AIMD Fast recovery (FR)

First two are compulsory in TCP, while the last one is recommendable

rate = CongWin

RTT Bytes/sec


TCP Slow Start

available bandwidth may be >> MSS/RTT

When connection begins, increase rate exponentially fast until first loss event

initial rate is slow but ramps up exponentially fast

Host A

one segment

RTT

Host B

time

two segments

four segments


Congestion avoidance

After 3 dup ACKs: CongWin is cut in half window then grows linearly

But after timeout event: CongWin instead set to 1 MSS; window then grows exponentially to a threshold, then grows linearly

3 dup ACKs indicates network capable of delivering some segments

timeout indicates a “more alarming” congestion scenario

Philosophy:


Summary: TCP Congestion Control

When CongWin is below Threshold, sender in slow-start phase, window grows exponentially.

When CongWin is above Threshold, sender is in congestion-avoidance phase, window grows linearly.

When a triple duplicate ACK occurs, Threshold set to CongWin/2 and CongWin set to Threshold.

When timeout occurs, Threshold set to CongWin/2 and CongWin is set to 1 MSS.

Note: TCP version varies the way congestion control is applied. For example, TCP Tahoe always reduces Congestion Window to its initial value after a loss event. Meanwhile, in TCP Reno, all the congestion phases take place.


Chapter 2 outline







TCP and UDP limitations TCP limitations:

TCP keeps strict order: head-of-line blocking may be a problem (data flow blocked until recovering a lost segment)

Byte-oriented nature of TCP: must use PSH to ensure data goes to app

No multi-home IP hosts Relatively vulnerable to some attacks (SYN

flooding)

UDP limitations Not reliable No data order No congestion control

Solution: SCTP


Stream Control Transport Protocol SCTP

Stream Control Transport Protocol, RFC 2960 Initially created by SIGTRAN group for PSTN

SCTP enhancements Multi-homed host support: redundancy & efficient flow Multiple flows in one association: solves head-of-line

blocking TCP problem

I P Network

Host A Host BApp 1

SCTP Port = 100

-------------------------------- OS

App 2SCTP Port = 200

-------------------------------- OS

NI-1 NI-3NI-2 NI

IP3 = 10.1.61.11

IP2 = 160.15.82.20

IP1 = 161.10.8.221

IP = 128.33.6.12


SCTP features SCTP features

Connection oriented Concepts

• Endpoints: SCTP endpoint is a list of addresses, same ports

• Association multiple possible IPs• {[160.15.82.20, 161.10.8.221, 10.1.61.11:100]}:

[128.33.6.12:200]} (two endpoints, one association – Figure -)

I P Network

Host A Host BApp 1

SCTP Port = 100

-------------------------------- OS

App 2SCTP Port = 200

-------------------------------- OS

NI-1 NI-3NI-2 NI

IP3 = 10.1.61.11

IP2 = 160.15.82.20

IP1 = 161.10.8.221

IP = 128.33.6.12


SCTP header

SCTP header First 32 bit words SCTP common header (similar to

UDP header) Verification tag: distinguish between associations;

prevent from attacks Chunks: building blocks


SCTP header SCTP chunks

Blocks with a 32-bit multiple length Different types

• Control (e.g., INIT, INIT-ACK, COOKIE-ECHO, COOKIE-ACK – four-way handshake -)

• Data (e.g. DATA)

transport layer advanced features1 chapter 2 transport layer advance features some of these slides...

Documents

transportlayer services2

tcp congestion control2

transport services

connectionoriented transport

transport protocol available

tcp segment structureurg

connection estabsetup

bidirectional data flow