1

Link Layer Switching

Connecting local networks Bridges Repeaters, Hubs, Bridges, Switches,

Routers, Gateways Virtual LANs

2

Ethernet

• 50 thick: 500 m• 50 thinn: 185 m

• max 4 repeaters• traffic on one segment means traffic on all other segments

3

CSMA/CD (IEEE 802.3)

A-MACPhys. A

B-MACPhys. B

C-MACPhys. C

Logical Link Control

Physical

Link

4

Bridges

Connection on link layer: forwarding based on MAC addresses self-learning bridges

operation Advantages and limitations

Spanning-tree bridges operation Advantages and limitations

5

Self-learning Bridge

MAC_1Phys_1

LLCMAC_1Phys_1

MAC_2Phys_2

LLCMAC_2Phys_2

Forwarder

Bridge

routing table

Network 1 Network 2

6

Self-learning Bridge

Driver interface 1. Driver interfaec 3. Driver interface 2.

LAN 1 LAN 2 LAN 3

Learning&

routing

Routing table

MAC-adr. Interface. timeMac-1- - - - -Mac-2

2- - 3

- -- -- -

 

7

Self-learning Bridge

ExtractSender and receiver MAC-adresser

Updateinterface #and timer

New entryMAC-addrinterface #and timer

Start

Senderknown?

Yes No

Look up inRouting table

Receiverknown?

Broadcast frame, except on receiving

interface

Put frame intocorrectoutgoing queue

End

Learning phase Forwarding phase

Look up inRouting table

8

Link Layer Switching

Multiple LANs connected by a backbone to handle a total load

higher than the capacity of a single LAN.

9

Bridge from 802.x to 802.y

IEEE 802 frame formats

10

Bridges from 802.x to 802.y

Operation of a LAN bridge from 802.11 to 802.3.

11

Local “Internetworking”

A configuration with four LANs and two bridges.

12

Problem with standard bridge

Two parallel transparent bridges.

13

Spanning tree

Goal: each bridge should identify the interfaes for forwarding traffic

Build a spanning tree From on root node

Self-configuring To all nodes

Only these interfaces in the spanning tree can forward traffic

Provides the shortest path for all traffic

14

Spanning Tree Algorithm

Configuration phase: Each nodes sends out:

Its own identity (ID) (MAC-address) ID to the root-bridge Number of hops to root-bridge

In this way, building up a spanning tree, bridge with lowest ID become root node

Start forwarding frames

15

Spanning Tree Bridges

(a) Interconnected LANs. (b) A spanning tree covering the LANs. The dotted lines are not part of the spanning tree.

16

Remote Bridges

Bridges can be used to connection physically distant local networks

17

Repeaters, Hubs, Bridges, Switches, Routers and Gateways

(a) Which device is in which layer.(b) Frames, packets, and headers.

18

Hub (Nav)

Hub

< 100 m

Hub

HubTransceiver

19

Repeaters, Hubs, Bridges, Switches, Routers and Gateways

(a) A hub. (b) A bridge. (c) a switch.

20

Switched Ethernet

Switch

Server Server

10, 100, 1000Mb/s

Switch:•Switches on MAC-addr•Buffers frames, therefor no collision•Competition only for switch capacity

21

Gigabit Ethernet

Gigabitswitch

Central server

Central server

Switch Switch Group serverGroup server

Working group 1 Working group 2

100/1000100/1000

1000 Mb/s100 Mb/s

22

Virtual LANsA building with centralized wiring using hubs and a switch.

23

Virtual LANs (2)

(a) Four physical LANs organized into two VLANs, gray and white, by two bridges. (b) The same 15 machines organized into two VLANs by switches.

24

The IEEE 802.1Q StandardTransition from legacy Ethernet to VLAN-aware Ethernet. The

shaded symbols are VLAN aware. The empty ones are not.

25

The IEEE 802.1Q Standard (2)

The 802.3 (legacy) and 802.1Q Ethernet frame formats.

26

Conclusion

Bridges: efficient connection alternative Limits/isolates collision domains Can be used for traffic isolation Do not consume IP addresses

Switches: High use degree, no danger of

collisions Used for establishing virtual LANs

27

Routing and Packet Switching

Goal Overview of how routing fits into the

Internet architecture Principles for typical routing protocols

Strengths and weaknesses Structure

Primary tasks of the network layer Datagram and virtual line Some performance considerations Routing and forwarding

28

Network layer

DiskDisk

Server Client

link

29

Tasks of the Network Layer

Responsible for end-to-end transport Addressing of machines Forwarding

Connectionless datagram; no fixed path through the network

Connection-oriented (e.g. MPLS or ATM) Three phases: connection establishment, data

transmission, teardown Fixed path through the network Relatively reliable and ordered transmission

Flow control

30

ForwardingR

B

A

R

LAN-A

LAN-B

31

Routing and lookup Mail: griff@ifi.uio.no Name to address conversion:

ifi.uio.no til IP address: 129.240.64.2 Find MAC-address to router and send

packet(s) Forward through the network w.r.t. the

network address Based on lookup in routing tables

At the destination router Convert machines IP address to a MAC

address Send packet to the receiving machine

32

Place of Routing in the architecture

Structured Network dimensioning

Where should lines be established? Capacity of lines

Traffic directioning Mapping of connections down to paths through

the net Routing to choose paths

Routing of individual packets Best effort

Routers choose the next hops separately for each packet

33

Routing

Routing tables can be computed based on state information about the network

Data exchanged between nodes: Between neighbour nodes (distance

vector routing; RIP) Between all nodes in the network (link

state routing; OSPF, IS-IS)

34

Routing types Static vs. dynamic

Dynamic with error handling, new links, changes of the load

Centralized vs. distributed Distributed when routes are computed at all

nodes Global vs. local topology knowledge Source routing vs. routing Kilde ruting vs. ruting

In source routing the source chooses the routing

In routing each router choose the next hop

35

Routing ParametersPerformance parameters•Number of hops•Price•Delay•capacity

Routing decisions made•In each node (distributed)•In a central node (routing center)•At the sender (source routing)

Sources of routing information•None•Local to the node•Neighbour nodes•Nodes along the path•All nodes in the network

Update interval•Continously•Periodic•In case of large load variations•In case of topology changes

36

Routing hierarchy In large networks

Hierarchically structured Link state

Open Shortest Path (OSPF) Intermediate System to Intermediate System

(IS-IS) On campus or in companies

Distance vector, RIP Static routing

Ad-hoc networks, stationary or mobile wireless networks

Many different protocols depending on scenarios

37

Router model

Forwardingprocess

Pre-process

In

1

2

3

1

2

3

Out

e

Principle structure of a router with three incoming and three outgoing connections

Routingtable

Topologydatabase

Routing prosess

Routecomputation

 

38

Routing alternatives Flooding Static routing Adaptive routing should handle

Loss of a link (error, e.g. cable is broken) Loss of a node (error, e.g. power loss, OS

crash) High traffic load (persistant of transient

congestion, bottleneck) Disadvantages

Complex, distributed, and not always correct Adaptivity must be balanced against additional

overhead Can lead to oscillations (route flapping) if

reactions are too fast Can be unattractive if reactions are too slow

39

Demands on a routing strategy

Shall give correct routes Shall demand minimal load on nodes Shall be stable and converge quickly Fair towards different data streams Provide optimal routes Scale with the size of the network Size with the number of destinations

40

Plug-and-play capabilities Find neighbour nodes and routers Detect when neighbours go up and

down Detect capacity of own links Send and receive topology

information Send after timer or major changes

to the network

41

Distance vector characteristics

Nodes exchange a vector with their shortest distance to all destinations Periodic exchange

Convergence is ensured

Advantage Simple

Disadvantages Vulnerable to errors Slow dissemination in case of problems

42

Distance Vector

AB C

D E

F1

2

5

22

9 1di1

. .diN

Di ==

si1

. .siN

Si ==

Distance vector Next node vectorDest. delay Next node A 0

- B 2 B C 5 C D 1

D E 6 C F 8 CNode A before change

53

44

Router model

Forwardingprocess

Pre-process

In

1

2

3

1

2

3

Out

e

Principle structure of a router with three incoming and three outgoing connections

Routingtable

Topologydatabase

Routing prosess

Routecomputation

 

45

Link state Routing database Routing table Periodical and in case of changes

Nodes flood their state onto the link to all other nodes At start, new nodes downlink the database from a

neighbour Different kinds of link

Point-to-point Point-to-multipoint Broadcast

Each node calculates the best route to all other nodes

Checkpoints Voting av entire database for link state at a sequence

number

46

LS routing protocol architecture

changechange

Protocol for handling of changes

Link state database

Routing algorithm

Rou

ting

tabl

e

route lookup

47

Flooding of link state Statistically reliable

Each node forwards on all interfaces All incoming link state packets

If sequence number of large than earlier sequence numbers Will most probably reach all node in the network

Content Sequence number

Avoid broadcast storms Node ID of the source

Topology Identify bi-directional links

List of all direct neighbour nodes with a cost function

Time-to-live

48

Link state, LSA

AB C

D E

F1

2

5

22

9 1

53

A B C D E F

A 2 5 1

B 2 2

C 5

D 1 2 9

E

F

Routin database i D

49

Link state problems/strengths Problems

Selection of a node that reports for a shared medium

Flooding does not scale for large networks

Division into hierarchical networks to limit flooding

Strengths All nodes have full topology knowledge Error have only local relevance

50

Link state problem

area 1 area 2

a

b

We have two problems with the link state method1. Static cost factor

• Can be the source of congestion, all traffic is routing through a single link2. Oscillation effects in forwarding traffic

• At one point in time a is the preferred router between areas• Then routing information is exchange• New tables are computed and b becomes the preferred router

51

Router and Routecenter

Network with router center

Route centerRouters do not have to participate in a routing protocol

Routing center receives status reports from routers

Transfers forwarding table to routers

52

Connection-oriented forwarding Establish a channel = a path through

the network Examples ATM, MPLS, X.25 Explicit signalling Data-driven signalling

Signalling protocol Routing protocol to choose the nodes

that should form the path In each node establishing a forwarding

table Incoming interface, channel – to outgoing

interface, channel

53

54

55

node x node y

C

1

1 2

V.C. tables

12

34

1 2

1

2 2a

b

c

b

V.C. tables

c

DA

B

a1 ; c ; 32 ; c ; 4

b1 ; c ; 22 ; c ; 1

1 ; b ; 22 ; b ; 13 ; a ; 14 ; a ; 2

c1 ; a; 32 ; a; 4

1 ; a; 12 ; a; 2

1 ; b ; 12 ; b ; 23 ; c ; 14 ; c ; 2

cba

a

 

Virtual lines

56

Dynamic cost in route computation

Adaptation of routes to load Move traffic to lines with lower load

Main problem Delay between measurement and computation Delay between route computation and traffic

arrival Fast variation in load

Bad predictability

Route flapping (oscillations) Overhead of exchanging the routing

information

57

Performance of the network

Performance of the networks means capacity, delay, delay variation (jitter), and reliable

Has several elements Transmission delay

Sending delay Signal propagiation time

Node delay Processing time Queueing time

58

Measuring the link stateNode N Node (N+1)

T0

T1

packetpacket

ACK

59

Topology example

(2)A

B C

DFE

G H

(7)

(2)(2)

(1)(6)

(4)

(2)

(3)(3)

(2)

Link 1

Link 2

Link 1

Link 2

Link 3

60

Shortest path tree for nodes

A

B

E

G

F

C

H

D

E B A

C

F

H

D

G

 

61

Routing tables for shortest path trees

Node A

Node Link no

1111111

BCDEFGH

Node E

Node Link no

1122232

ABCDFGH

62

Packet size and delay

63

Modified load variation

1

2

3

4

5

Kost

0,51,0

Utnyttelsesgrad

Jordbunden

Satellitt

Køteoretiskforsinkelse

 

64

Timing in line and packet switching