lis.gif

2.gif1.gif

Creating Larger Networks

 

There are physical limits to how big a network can become, for example

tball.gif   The longest single wire in a 10BaseT network (UTP) cannot be more than    100 meters.

tball.gif   If too many computers try to use a LAN, performance may become               unacceptable.

This chapter deals with solving these problems.

Repeaters

A repeater is like a two-way digital amplifier. It has two cable connections sockets. Whatever is received into one connection is re-transmitted through the other connection stronger and cleaner.

Repeaters work at Layer 1, the Physical Layer, of the OSI model. They are dumb - they have no intelligence. They do not understand the meaning of the data they pass, they simply re-generate it, bit a at time, and pass it on. This means that the segments on each side of the repeater must be using exactly the same protocols. A repeater cannot, for example, be used to join an Ethernet with a Token Ring network.

It is possible to buy repeaters that accept two different media types. For example one socket could be for Thinnet and the other twisted pair.

The simplest kind of hub is just a ‘multi-port’ repeater. It has many sockets. Whatever is received into one socket is regenerated and re-transmitted out of each of the other sockets.

Bridges

Bridges are similar to repeaters but much more sophisticated. They work at Layer 2, the Data Link Layer of the OSI model. A bridge is really a special-purpose computer. It has intelligence.

 

Consider the following diagram.

l25p1.gif1. In this simple example, the bridge divides the network into two halves, called segments.

2. Every NIC on the network has a hardware address - 1 through 8 in the diagram.

3. Every packet of data put onto the media contains both a source address and a destination address.

4. Imagine that the first packet to be sent is from #1 and is intended for #4. The hub of segment A will send this message to #2, #3, #4 and to the bridge.

5. Because the bridge is intelligent it will learn from this. It will learn that the sender, #1, is on segment A. It will then start to build a routing table in its RAM:

 

Address

Which Segment?

 

#1

A

 

Because the bridge has not learned which segment the destination, #4, is on, it will send the packet to segment B - just in case - where it will be ignored.

6. Imagine that the next packet to be sent is from #6 and is intended for #2.

The bridge does not yet have an entry in its routing table for #6, so it will add one.

 

Address

Which Segment?

 

#1

A

 

#6

B

 

Because it does not yet know which segment the destination, #2, is on, the bridge will forward it to the other segment.

7. The next packet is from #4 to #1. The bridge can tell that the sender, #4, is on segment A (because the packet arrives on the wire from segment A). The bridge has already learned that the destination, #1, is also on segment A. In this case the bridge does not pass the packet to segment B.

It does, however, add information about #4 to the routing table:

 

Address

Which Segment?

 

#1

A

 

#4

A

 

#6

B

 

8. The next packet is from #3 to #6. The bridge learns that #3 is on segment A. It adds this to its routing table.

 

Address

Which Segment?

 

#1

A

 

#3

A

 

#4

A

 

#6

B

 

It already know that #6 is on segment B, so it forwards the packet to segment B.

9. As more packets are sent the bridge continues to build its routing table until it is complete.

In the above there was only one bridge and two segments. In a big network, there may be several bridges and many segments.

l25p2.gif

 

Here a bridge connects three segments. It looks at the destination NIC address of every packet it receives. If it knows which segment that NIC is one, it forwards the packet to that segment only. If it does not know where the NIC is it forwards to packet to all segments except the one it came from.

 

The advantage of a bridge when compared to a repeater is that network traffic is reduced. Packets stay within their segment unless it is necessary to move to another.

Because bridges work at the Data Link Layer of the OSI model, they understand cable access methods. Because of this they can translate between access methods. So a bridge may be used to link an Ethernet segment with a Token Ring segment.

l25p3.gif

When a packet must be passed from the Ethernet to the Token Ring, the bridge strips off the Ethernet control bits and replaces them with Token Ring information. We say that the bridge regenerates at the packet level.

Routers

A router is like a bridge, but it works at the next layer up the protocol stack - layer 3, the Network Layer.

Instead of looking at the NIC address of packets it looks at the destination network address - added by the Network Layer. Because of this a router can only work with routable protocols - ones that add destination network addresses to packets. (Novell’s SPX/IPX and TCP/IP are common routable protocols.)

Every data packet transmitted using a routable protocol has two parts to its source and destination addresses.

1. A number indicating which network.

2. A number indicating which station on that network.

Routers work only with the network ID - they do not worry about the individual station ID.

Routers are not connected directly to work-stations. Instead they are connected to a hub locally, and to other routers. When a router receives a packet from one of its local machines it looks only at the ID number of the destination network:

1. If the destination network is the local network, then it does not forward the packet.

2. Otherwise, if it knows the route to the destination network, it sends the packet to the router dealing with that network.

3. Otherwise, it sends the packet to a default router and lets that router figure out what to do with the packet.

 

Consider this network:

l25p4.gif

Imagine that a station on segment A wishes to send a packet to a machine on segment B. The packet goes from its source, to the hub in segment A and so to Router A. The router looks only at the destination network ID, and sees that the packet must go to segment B. The packet is then passed to the router for segment B.

The routers can also cope with multi-hop routes between source and destination. For example for a station connected to Router A to send a packet to a station connected to Router C, the packet must pass through another router on the way. Software inside each router builds up a table giving routes to all other routers on the network.

Routers can also handle multiple paths between networks. For example, in the diagram Router A can talk to Router C in two ways - either via B or via E and D. Some routers can choose dynamically which path is likely to be best, by considering the volume of network traffic along each route. Should Router B fail, the other routers will be smart enough to choose alternative paths. Where long distance leased lines are used between some points, routers in the system can be programmed with usage costs and will automatically select the cheapest route.

Routers can also be programmed to act as filters - passing some packets but blocking others. For example, a router could be used to allow teachers to access the student network, while preventing students accessing the teacher’s network.

Note: Using routers each network segment can use a different low level protocol as long as they share a common, routable, Network Layer protocol - such as TCP/IP.

For example, the segment connected to Router A could be running TCP/IP on an Ethernet while the segment connected to Router B could be running TCP/IP on a Token Ring network.

Brouters

A brouter is a cross between a bridge and a router. It is a single device that:

tball.gif   Routes packets addressed using a routable protocol.

tball.gif   Acts as a bridge for packets addressed using non-routable protocols.

Gateways

A gateway is a server running software, and probably with special hardware, which enables a PC LAN to swap data with a completely different system - for example a mainframe computer.

The gateway takes a packet and strips off all of the protocol information used by the LAN. to leave just the data. It then wraps the data in whatever protocol is needed to talk to the other system.

l25p5.gif

Summary

Repeater

Layer 1 - Physical Layer.

tball.gif   Dumb.

tball.gif   Regenerates incoming bits and sends them out on all other wires.

Bridge

Layer 2 - Data Link Layer.

tball.gif   Builds routing table.

tball.gif   Used to divide the network into segments.

tball.gif   Looks at address of the destination NIC. If the location of the destination is known, then the packet is sent only to the correct segment. If the location is not know the packet is sent to all segments except the source segment.

tball.gif   Can join segments with different low level protocols - e.g. Ethernet with Token Ring.

 

                          Router

Layer 3 - Network Layer.

tball.gif   Works only with routable protocols.

tball.gif   Routers communicate with other routers.

tball.gif   Routes packets based on the network ID part of the destination address.

tball.gif   Can handle ‘multi-hop’ (A-B-C) routing.

tball.gif   Can select best route when several exist.

tball.gif   Can join segments with different low level protocols - e.g. Ethernet with Token Ring.

 

                           Gateway

Joins a LAN to totally dissimilar system - e.g. to a mainframe.

line_mou.gif

LISTEN.gif

Back to top

arrow003.gifarrow002.gif

| list of lessons | Course Plan | Instructor's Info | Links | UAE | Guestbook |View Guestbook | Books | Search|HCT|

home.gif