chapter7 l1

38
1 Chapter 7—packet- switching Chapter 5: data-link layer protocols (some protocols are nearly same for transport layer) Chapter 6: MAC sublayer (for broadcast networks). This chapter: network layer Telephony networks (i.e., circuit-switching) for stream of real time voice Packet-switching networks for short messages, burst information, as well as real-time applications.

Upload: chandanv8

Post on 13-Dec-2014

848 views

Category:

Documents


0 download

DESCRIPTION

 

TRANSCRIPT

Page 1: Chapter7 l1

1

Chapter 7—packet-switching

• Chapter 5: data-link layer protocols (some protocols are nearly same for transport layer)

• Chapter 6: MAC sublayer (for broadcast networks).

• This chapter: network layer– Telephony networks (i.e., circuit-switching) for stream

of real time voice

– Packet-switching networks for short messages, burst information, as well as real-time applications.

Page 2: Chapter7 l1

2

Two views of networks• External view: what services provided to transport layer by

a network (i.e. network layer)– Need connection setup or not– What QoSs are provided– Services should be independent from underlying networks so that

transport layer can run over any networks as long as the networks provide the services

• Internal view: – physical topology,– datagram message transfer or virtual circuit information transfer, – addressing and routing, congestion control.

• In this chapter, networks indicate systems to transfer information from one end to the other. It is reasonable to consider networks to be network layer here.

Page 3: Chapter7 l1

3

Comparisons of two views by examples

• Broadcast networks and packet-switched networks• From external view, both networks provides

transfer of information between users, not too much different

• But from internal view, very different:– A broadcast network (such as LANs) is small, addressing

is simple, frame transferred in one hop so no routing is needed

– In a packet-switching network, addressing must accommodate large-scale networks and routing is necessary.

Page 4: Chapter7 l1

4

What we will discuss

• Network services and internal network operations• Physical view of networks,• Datagram and virtual circuit• Routing• Shortest path algorithms• ATM networks• Traffic management and congestion control (if

have time)

Page 5: Chapter7 l1

5

Figure 7.1

t0t1

Network

Transfer of a block of info. VS. a sequence of blocks

Page 6: Chapter7 l1

6

Services provided by network layer

• Two basic kinds of services– One view: Transfer of a block of information VS. a sequence of blocks – Second view: connectionless VS. connection-oriented.– IP provides connectionless service VS. ATM provides connection-oriented service

• Other services such as:– Best-offer connectionless services– Low-delay connectionless service– Connection-oriented reliable stream service– Connection-oriented transfer of packets with delay and bandwidth guarantees

• End-to-end argument:– Functions should be placed as close as possible to the application since the application is in best

position to determine whether a function is being carried out completely and correctly,– Therefore, network layer should provide minimum functions required to meet application

requirements and performance. Leave more functions to upper layers such as transport layer.

Page 7: Chapter7 l1

7

Figure 7.2

Physicallayer

Data linklayer

Physicallayer

Data linklayer

End system

Networklayer

Networklayer

Physicallayer

Data linklayer

Networklayer

Physicallayer

Data linklayer

Networklayer

Transportlayer

Transportlayer

MessagesMessages

Segments

End system

Networkservice

Networkservice

protocol stack: network provides minimum required services to transport layer

Page 8: Chapter7 l1

8

Discussion of connectionless vs. connection-oriented

Both connection-oriented and connectionless services in the upper layer can be implemented over a connectionless network (layer) (such as IP), similarly, can be implemented over a connection-oriented network (layer) (such as ATM).

In network layer, connectionless service is also called datagram service, connection-oriented service called virtual-circuit service.

Which is preferred?From end-to-end argument, connectionless network (IP) is preferred because

IP makes intermediate nodes as simple as possible and put the burden on endsystems. Moreover, it makes the network grow easily (more scalable).

Why ATM which is connection-oriented?Main reason is QoS because connection-oriented network can easily guarantee bandwidth, delay, etc. Detail motivation behind ATM will be discussed later.

Network layer: connectionless(IP) connection-oriented (ATM)

Transport layer: connectionless connection-oriented

Thinking: differences (or implications) of connection-oriented in transport layer and in network layer. What the connection between two ends in transport layer means?

Page 9: Chapter7 l1

9

Most basic functions of a network (layer)

Routing and forwarding.

Priority and scheduling (to provide QoS)

Congestion control (also in transport layer)

Segmentation (to deal with different frame sizes of underlying systems)Alternatively, network (layer) sends error message to let edge system to do segmentation.

Addressing 1. to deal with different address formats when interconnecting networks.

2. Hierarchical address for scalability.

Page 10: Chapter7 l1

10

Figure 7.4

.

.

.MUX

Network access

Node

Packet network topology --Access multiplexer

Multiplexer combines multiple bursty flowsinto one aggregated flow.

Access node forwardspackets into backbonepacket networks.

Page 11: Chapter7 l1

11

LAN

Bridge

LAN 1

LAN 2

(a) (b)

Figure 7.5

Packet network topology --LAN

Now nearly every LAN is connected to packet-switching network (the Internet) to share information globally and to make the Internet extremely large.

LANs were introduced for sharing of resources in a organization.They are basic building components for Wide Area Networks.

Page 12: Chapter7 l1

12

RR

RR

S

SS

s

s s

s

ss

s

ss

s

R

s

R

Backbone

To internet or wide area network

Organization Servers

Gateway

Departmental Server

Figure 7.6

Campus network

LANsextended LANs subnetworks (by backbone networks) campus network university network, finally connect to Internet,

bridges, routers and gateway,… play key-role in networks

Backnone network may be connected by links, LAN, or ATMs.

Servers provide various services.

Page 13: Chapter7 l1

13

Interdomain level

Intradomain level

LAN level

Autonomous systemor domain

Border routers

Border routers

Figure 7.7

Internet service provider

Intradomain and interdomain levels

Domain: indicate the routers run the same routing protocols.

Autonomous systems: includeddomains under a single administrations

•Various networks are connected through ISP.

•Tremendous small office and home (SOHO) users connect to the Internet through ISP, using dynamic IPs.

Page 14: Chapter7 l1

14

RA

RB

RC

Route server

NAP

National service provider A

National service provider B

National service provider C

LAN

NAPNAP

(a)

(b)

Figure 7.8

National service providers are connected by NAPs (national access point)Route server distributes routing information to routers.

Page 15: Chapter7 l1

15

Packet-switching topology• What a big picture of topology for picket-switching

networks!!!– Hierarchical structure, from LANs campus, university

and organization networksInternet

– ISPs provide backbone networks to connect tremendous different networks and enormous home PCs

– Switches (bridges, routers, gateways): key elements

– Various servers provides rich services to users

– Domain, intradomain, interdomain for easy administration and network management.

Page 16: Chapter7 l1

16

Network

Packet switch

Transmission link

Figure 7.9

Simplified picture of switched networks

1. Transmission lines and packet switches2. They provide connectivity between any source and destination dynamically

3. The resources are allocated when needed, so can be shared among multiple users 4. Information can be transferred in connection-oriented or connectionless manner.

Page 17: Chapter7 l1

17

Brief discussion of switches

• Switches are generic term for switching devices in switched networks– Circuit-switches in telephony network– LAN switches in LAN connection, i.e. bridges,

Ethernet switch,… – Packet switches in pack switching networks,

i.e. routers and gateways.

Page 18: Chapter7 l1

18

Control

1

2

3

N

Line Card

Line Card

Line Card

Line CardIn

terc

onne

ctio

nFa

bric

Line Card

Line Card

Line Card

Line Card

1

2

3

N

Figure 7.10

…… ……

Structure of a packet switch/router

Components: input ports, output ports, interconnection fabric, controller

A line card handles several pairs of input/output ports and implements physical layer and data link layer functions: symbol timing, line coding, framing, physical addressing, error checking,MAC protocol, data link protocol, and buffering (speed mismatch between lines and fabric).

Page 19: Chapter7 l1

19

CPU

1

2

3

N

NIC Card

NIC Card

NIC Card

NIC Card Mai

n M

emor

y

I/OBus

Figure 7.11

……

A PC or workstation can be configured to become a switch

1.Several NICs are inserted into expansion slots, 2. Install routing and other protocols in the computer

Operations:1. NIC gets frames from networks and de-encapsulate packets2. Packets are transferred into memory using I/O bus3. CPU performs required routing and protocol processing4. Packets are transferred from memory to an appropriate NIC

5. The NIC forms new frame and sends out

Page 20: Chapter7 l1

20

Three basic resources and bottlenecks in switches

• Processing, memory and bus bandwidth– Processing implements protocols, hence processing

capacity places a limit on maximum rate at which switch can operate

– Memory stores packets, hence the amount of memory determines the rate at which packets are lost, placing another limit on switch load, moreover memory bandwidth also place limit on switch rate

– I/O bus bandwidth places a limit on total rate at which information can be transferred between ports.

Page 21: Chapter7 l1

21

1

2

N

1

2

N

Figure 7.12

……

Input port demultiplexes incoming packet stream; packets are routed to output port; output port multiuplexes outgoing packet stream

Switches/routers play a key role in controlling packet flows, thusefficiently using network bandwidth and optimizing performance.

Page 22: Chapter7 l1

22

Networknodes

Message

SubscriberB

SubscriberA

Message

Message

Message

Figure 7.13

Connectionless packet switching—originated from message switching

•message has header with source & destination address• CRC check bit are used to detect errors

•Each switch check error, if yes, ask retransmission, if not find next hop.•Message enter into a QUEUE to wait for line free to transmit•Increased utilization of line is at the expense of queuing delay•Loss of message may occur because of insufficient buffer.•End-to-end error recovery is needed

Page 23: Chapter7 l1

23

t

t

t

t

Delay

Source

Destination

T

p

Minimum Delay = 3p + 3T, ( remember it,will use it later)

Switch 1

Switch 2

Figure 7.14

Delays in message switching

Back

p: propagation delayT: message transmission time

Page 24: Chapter7 l1

24

Packet switching VS. Message switching

Why packet switching?

1. Long message has more chance to incur error than short packet

2. Long message is not suitable for interactive application becauseit causes very long waiting delay on other messages. Moreoverthe delay for one message is generally longer than the total delayof packets the message is divided into.

Examples: L=1M=106 bits over two hops. Bit error rate: p=10-6

the probability without error: Pc=(1-p)L=(1-10-6)106eLp=e-1 1/3.So on average, 1/(1/3)=3 tries for 1st hop. Similarly, 3 tries for 2nd hop, Total six tries, 6M.Suppose divided into 105 bit packets, the no-error probability 0.9.So each packet needs 1/0.9=1.1 times on average. So entire messageGet transmitted over each hop using 1.1M bits transmission.So total 2.2M bit transmission over two hops.

Example: come soon.

Page 25: Chapter7 l1

25

Packet 2

Packet 1

Packet 1

Packet 2

Packet 2

Figure 7.15

Datagram packet switching

Operations: 1. Address information is included in header 2. CRC for error recovery 3. switches inspect destination address in header to determine next hop 4. Packets are put in QUEUE to wait for line becoming available 5. Sharing lines among multiple packets, high utilization is at the expense of queue delay 6. Packets travel independently and may along different paths

7. Route may be detoured,thus bypassing failure and congestion8. Packets may arrive out of order, resequencing may be required

Page 26: Chapter7 l1

26

t

t

t

t

31 2

31 2

321

3p + 2(T/3) first bit received

3p + 3(T/3) first bit released

3p + 5 (T/3) last bit released

Lp + (L-1)P first bit received

Lp + LP first bit released

Lp + LP + (k-1)P last bit releasedwhere T = k P

3 hops L hops

p

p + P

p + P

Source

Destination

Switch 1

Switch 2

Figure 7.16

Delay in packet switching networks

Assume three packets along the same path and transmitted in succession

Page 27: Chapter7 l1

27

Delay comparison of message switch vs. packet switch

• Support two switches and 3 packets /message– p: propagation delay, T: message transmission time, P: packet

transmission time, and T=3P– Delay in message switching: 3p+3T see detail– Delay in packet switching: 3p+T+2T/3

• Support L-1 switches and k packets/message:– T=kP– Delay in message switching: Lp+LT =Lp+LkP– Delay in packet switching (k packets): Lp+LP+(k-1)P

• Results: message switching involves additional delay of (L-1)(k-1)P.

• Note, in the above, we neglect queuing and processing at each switch.

Page 28: Chapter7 l1

28

Destinationaddress

Next hop

1345 2343

2458

32450785

3784

7612

1566

Figure 7.16

Routing table in connectionless packet switching

Designing routing table is a key issue in packet-switching networksWhich requires: knowledge of network topology and traffic levels.Moreover, size of table will become very large when network size increase.

Page 29: Chapter7 l1

29

Example –IP internetworks

• Objective of IP is to provide connectionless transfer service across heterogeneous networks

• Read it through textbook.

Page 30: Chapter7 l1

30

Virtual-circuit packet switching• Establishment of connection between source &

destination prior to the transfer of packets See figure• Connect request is sent by source and connect

confirm is replied by destination See figure • Signal exchanges is performed in virtual-circuit setup

See figure• All packets from a connection follow the same path.• Admission control could be used to limit load on

networks or specific links• Routing is easier once connection was setup routing

table

Page 31: Chapter7 l1

31

Packet

Packet

Figure 7.17

back

Virtual-circuit switching

Question: why called virtual-circuit?

Because resources, e.g., switches, buffers and lines, are shared by packets from multiple connections, not dedicated to one specific connection.

Page 32: Chapter7 l1

32

t

t

t

t

31 2

31 2

321

Release

Connect request

CR

CR Connect confirm

CC

CC

Back

Beginning latency and delay of packets in virtual-circuit packet switching

Page 33: Chapter7 l1

33

SW 1

SW 2

SW n

Connect request

Connect request

Connect request

Connect confirm

Connect confirm

Figure 7.20

Back

Signaling message exchanges in virtual-circuit setup

Parameters such as buffer, bandwidth, delay requirements were set in every switch along the path during setup

Page 34: Chapter7 l1

34

Identifier Outputport

15 15

58

13

13

7

27

12

Nextidentifier

44

23

16

34

Entry for packetswith identifier 15

Figure 7.21

Example of virtual-circuit routing table for an input port

Each connection is identified by VCI (virtual-circuit identifier). VCI may be different along the path for the same connection: the input VCI in a switch is different from output VCI for the same connection, called local VCIs.VCI is in the header of a packet, which is much smaller than IP address, as in IP protocol.Table lookup will find the output port # and output VCI based on input VCI, which is faster.

Page 35: Chapter7 l1

35

Pro and con about virtual-circuit switching• Pro:

– Header is shorter, resources allocated during setup, admission and congestion are easier.

– Minimum delay reduces further. See figure

• Con:– Every router needs to maintain state information

about all connections– Once there is a failure, all affected connections

must be set up again.

• Example –ATM networks

Page 36: Chapter7 l1

36

31 2

31 2

321

Minimum Delay = 3p+T t

t

t

tSource

Destination

Switch 1

Switch 2

Figure 7.22

Cut-through packet switching

Forward a packet as soon as the header is received and table lookup is carries out.

Assumptions: no error check, all lines are available.

Back

Page 37: Chapter7 l1

37

t

t

t

t

31 2

31 2

321

3p + 2(T/3) first bit received

3p + 3(T/3) first bit released

3p + 5 (T/3) last bit released

Lp + (L-1)P first bit received

Lp + LP first bit released

Lp + LP + (k-1)P last bit releasedwhere T = k P

3 hops L hops

p

p + P

p + P

Source

Destination

Switch 1

Switch 2

Figure 7.16

Delay in packet switching networks

Assume three packets along the same path and transmitted in succession

Page 38: Chapter7 l1

38

Example –ATM networks• Connection-oriented networks• All information must be divided into fixed-length

very small packets called cell (53 bytes). (Why?)• The setup gives a chance for negotiating parameters

between user requirement and network commitment, resources are allocated along the path

• The connection is defined by a chain of local identifiers VCIs.

• Assume low-error rate optical channel, so error control is done only end-to-end

• Is intended to support a large range of applications