Internet Working

Internetworking To make the computer networks more useful andsuitable for many emerging applications, it is necessaryto connect individual heterogeneous networks, bothLAN and WAN, distributed across the world usingsuitable hardware and software in such a way that itgives the user the illusion of a single network.

Basic motivations behind internetworking are asfollows:

To provide a link between networks.

To provide a route for delivery of data betweenprocesses on different networks.

To accommodate a number of differences among thenetworks:Addressing scheme (Flat versus hierarchical(IP))Maximum packet sizeQoS(present or absent)Routing techniques (unicasting,multicasting)Protocols(IP,IPX,SNA,ATM etc.)Connection oriented/connectionless services

This module is concerned with internetworking, whichallows communication across isolated networkboundaries in a seamless manner.

Why interconnect networks? Reliability Performance Security

Backbone Network Components Two basic components to the BN

hardware devices that connect the networks to the backbone

repeaters hubs bridges switches routers brouters gateways

network cable

Repeaters A single Ethernet segment can have a maximum length of 500 meters with a maximum of 100 stations (in a cheapernet segment it is 185m). To extend the length of the network, a repeater may be used as shown in Fig.Ethernet allows five segments to be used in cascade to have a maximum network span of 2.5 km.With reference of the ISO model, a repeater is considered as a level-1 relay as depicted in Fig. It simply repeats, retimes and amplifies the bits it receives. The repeater is merely used to extend the span of a single LAN.

Important features of a repeater are as follows:

A repeater connects different segments of a LAN A repeater forwards every frame it receives A repeater is a regenerator, not an amplifier It can be used to create a single extended LAN

Hubs Hub is a generic term, but commonly refers to a multiport

repeater. (Active Hubs) It can be used to create multiple levels of hierarchy of

stations. The stations connect to the hub with RJ-45 connector

having maximum segment length is 100 meters. This type of interconnected set of stations is easy to

maintain and diagnose. A passive hub is just a connector. It just a point where the signals coming from different

stations collide;the hub is a collision point and placed below the physical layer.

Hubs very simple devices that pass all traffic in

both directions between the LAN sections they link

same or different cable types use physical layer protocols pass on every message used to connect LANs of similar technology,

or to extend the distance of one LAN can be called repeaters or amplifiers

HUB Devices

HUB (MAU)

Repeater/Amplifier

Hubs inexpensive easy to Install can connect different media very little delay

limited distance between devices limited on the number of repeaters no protocol or rate conversion no error detection does not filter

Bridges A bridge operates both in physical and data-link layer

A bridge uses a table for filtering/routing

A bridge does not change the physical (MAC) addresses in a frame

connect two LAN segments that use the same data link and network protocol

same or different cable types

forward only those messages that need to go out (filtering)

learn whether to forward packets

combination of black box hardware and software

BridgesUse of bridges offer a number of advantages, such as higher reliability performance security convenience larger geographic coverage

Types of bridges: Transparent Bridges Source routing bridges

Bridges Interconnecting

HUB (MAU)

Repeater/Amplifier

HUB (MAU)

Repeater/Amplifier

Bridge

Transparent BridgesThe transparent bridge uses two processes known as bridge forwarding and bridge learning.

It performs two functions: Forwarding of framesLearning to create the forwarding table

If the destination address is present in the forwarding database already created, the packet is forwarded to the port number to which the destination host is attached. If it is not present, forwarding is done on all parts (flooding). This process is known as bridge forwarding.

Moreover, as each frame arrives, its source address indicates where a particular host is situated, so that the bridge learns which way to forward frames to that address. This process is known as bridge learning.

Bridge ForwardingBridge forwarding operation is

Discard the frame if source and destination addresses are same Forward the frame if the source and destination addresses are different and destination address is present in the table Use flooding if destination address is not present in the table

Bridge Learning

At the time of installation of a transparent bridge, the database in the form of a table is empty.

As a packet is encountered, the bridge checks its source address and build up a table by associating a source address with a port address to which it is connected.

Creation of a bridge-forwarding table

Loop Problem Forwarding and learning processes work without any problem as

long as there is no redundant bridge in the system. On the other hand, redundancy is desirable from the viewpoint of

reliability, so that the function of a failed bridge is taken over by a redundant bridge.

The existence of redundant bridges creates the so-called loop problem as illustrated with the help of Fig.

Assuming that after initialization tables in both the bridges are empty let us consider the following steps:

Step 1. Station-A sends a frame to Station-B. Both the bridges forward the frame to LAN Y and update the table with the source address of A.

Step 2. Now there are two copies of the frame on LAN-Y. The copy sent by Bridge-a is received by Bridge-b and vice versa. As both the bridges have no information about Station B, both will forward the frames to LAN-X.

Loop Problem Step 3. Again both the bridges will forward the frames to

LAN-Y because of the lack of information of the Station B in their database and again Step-2 will be repeated, and so on.

So, the frame will continue to loop around the two LANs indefinitely.

Spanning Tree As redundancy creates loop problem in the system, it is very

undesirable. To prevent loop problem and proper working ofthe forwarding and learning processes, there must be onlyone path between any pair of bridges and LANs betweenany two segments in the entire bridged LAN. The IEEEspecification requires that the bridges use a specialtopology. Such a topology is known as spanning tree (agraph where there is no loop) topology.

Without changing the physical topology, a logical topology iscreated that overlay on the physical one by using thefollowing steps:

Select a bridge as Root-bridge, which has the smallest ID. Select Root ports for all the bridges, except for the root

bridge, which has least-cost path (say minimum number ofhops) to the root bridge.

Choose a Designated bridge, which has least-cost path tothe Root-bridge, in each LAN.

Select a port as Designated port that gives least-cost path from the Designated bridge to the Root bridge.

Mark the designated port and the root ports as Forwarding ports and the remaining ones as Blocking ports.

The spanning tree of a network of bridges is shown in Fig. The forwarding ports are shown as solid lines, whereas the blocked ports are shown as dotted lines.

Source Routing Bridges A transparent bridges duties include filtering frames,

forwarding and blocking In a source routing, a sending station defines the bridges

that frame must visit. The addresses of these bridges are included in the

frame. The source gets these bridges addresses through the

exchange of special frames called discovery frames with the destination prior to sending the data frame.

These bridges were designed by IEEE to be used with Token Ring LANs.

Bridges may be different data rates and different

media easy to Install can learn the ports for data transmission

understand only data link layer protocols and addresses

no protocol conversion broadcasts when it does not know the

address

Switches connect more than two LAN segments that

use the same data link and network protocol.

operate at the data link layer same or different type cable ports are usually provided for 4, 8, 16, 32

,64,128 or 256 LAN segments ports are used simultaneously connect lower speed segments to high

speed BN

Switches Cut-through switches

use circuit-switching to immediately connect the port with the incoming message to the correct outgoing port

very fast as decisions are done in hardware outgoing packet is lost if port is in use

Store-and-forward switches copy the incoming packet to memory prior to

processing the destination address -- transmit it when the outgoing port is ready

Switches Interconnecting

Wing C Wing C

Wing A Wing B

First Floor Switch

Switches much more sophisticated than previously enable all ports to work at the same time can convert protocols configurable high speed

understand only data link layer protocols and addresses

much more expensive then previous options higher maintenance

Routers connect two or more LANs that use the same

or different data link protocols, but the same network protocol.

same or different cable types operate at the network layer forward only messages that need to go out routers use the internetwork address internal routing tables only processes messages addressed to it

Routers choose the best route to send the packet

(path) IDs of other networks paths to the networks relative efficiency of the paths

Routers The router must deal with network

differences:

addressing schemes minimum packet size interfaces reliability

\

Routers InterconnectingRouter

EthernetLAN2

Token RingLAN1

X.25 Networkthe cloud

Routers can mix-in-match protocols and convert

them enable all ports to work at the same time can be used as an extra layer of security configurable high speed

hard to configure and manage access lists must be kept current high maintenance/high training costs very expensive

Brouters devices that combine the functions of both

bridges and routers operate at both the data link and network

layers same or different data link protocol same network protocol as fast as bridges for same data link type

networks

Gateways complex machines that are interfaces between

two or more dissimilar networks connect two or more LANs that use the same

or different data link layer, network layer, and cable types

operates at the network layer (3) or higher layers (4-7)

forwards only those messages that need to go out

a combination of both hardware and software translates one network protocol to another translates data formats

Hub Physical All transferred S/D Same SameBridge Data link Filtered using S/D Same Same

data link layer add. Switch Data link Switched using S/D Same Same

data link layer add.Router Network Routed using S/D S/D Same

network layer add.Brouter Data link & Filtered & routed S/D S/D Same

NetworkGateway Network Routed using S/D S/D S/D

network layer add.

Physical Data Link NetworkDevice Operates at Messages Layer Layer Layer

Backbone Network Devices

Terminology Warnings Multiprotocol bridges translate between

different data link layer protocols. Multiprotocol routers translate between

different network layer protocols. Protocol filtering bridges forward only packets

of a certain type, i.e., token-ring or ethernet Encapsulating bridges connect networks with

different data link protocols, encapsulating messages with correct protocol for transmission

Layer-3 switches (IP switches) - can also switch messages based on their network layer address