Take the 2-minute tour ×
Super User is a question and answer site for computer enthusiasts and power users. It's 100% free, no registration required.

From Networking all in one For Dummies:

"The second reason for subnetting is that even if a single organization has
thousands of network devices, operating all those devices with the same
network ID would slow the network to a crawl. The way TCP/IP works
dictates that all the computers with the same network ID must be on the
same physical network. The physical network comprises a single broadcast
domain, which means that a single network medium must carry all the traffic
for the network."

operating all those devices with the same network ID would slow the network to a crawl.?

same network ID must be on the same physical network. = The physical network comprises a single broadcast domain?

Sharing network ID can slow a network? Is this always true? I think this is only true for hubbed networks. Would using switches to connect the hosts prevent this?

Can anyone politely expand on this?

share|improve this question

2 Answers 2

up vote 2 down vote accepted

The example above is simply an example. The reality is that any network structure can become saturated given enough data transfer from enough hosts that the network fabric cannot move the data quickly enough.

Switches are the primary network device in use today, when you don't see the type of device defined you can safely assume it is a switch rather than a hub. The advantage of a switch over a hub is that it will not forward a packet to all ports if it has seen the destination mac address previously on one of its ports. Then the packet will only be forwarded out of that port.

This is great provided the mac has been seen and isn't a broadcast packet. Where this is not the case, the switch must forward the packet to all ports. There will always be some broadcast traffic on a network, and it is dependent on what is going on the network as to when you get to the point where the traffic starts to saturate the switches. No matter what percentage of traffic is broadcast, there will be capacity limits in the switch itself.

At this point, you have two options - segregate your network, or increase the capacity of your switch fabric.

In the vast majority of cases, segregating the network is the logical and sensible approach.

share|improve this answer
    
When you said that there will always be some broadcast traffic on a network, did you mean they are the result of the communications between the computers on that network? (done by protocol modules, perhaps?) –  WikiWitz Apr 6 '12 at 15:01
    
Some protocols are broadcast only, for sure, or have a broadcast component. But even in the simplest IP network, arp will be running, which is broadcast as is DHCP. But the other point is important too - any traffic that isn't destined for a learned mac address by a switch has to be broadcast. –  Paul Apr 6 '12 at 22:51

The physical network comprises a single broadcast domain, which means that a single network medium must carry all the traffic for the network.

This is flat out wrong. A broadcast domain simply means that any broadcast from any node on the network goes to all of the others. It has nothing to do with the type or number of media. The more nodes in the broadcast domain, the more broadcast traffic there is, so splitting up the network into separate segments can cut that down, but broadcast traffic is normally only a tiny fraction of the total traffic on a network, so this is rarely done.

share|improve this answer

Your Answer

 
discard

By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.