1

If the speed of the CSMA/CD network is scaled up and if all the other parameters stay unchanged does that mean the utilization will stay unchanged?

  • 2
    Could you rephrase the question? As is very unclear. Network utilization is the ratio of current network traffic to the maximum traffic that the port can handle. What do you mean by scaling up? – Datagram.Network Feb 13 '17 at 12:56
  • Did any answer help you? If so, you should accept the answer so that the question doesn't keep popping up forever, looking for an answer. Alternatively, you could provide and accept your own answer. – Ron Maupin Aug 16 '17 at 22:57
2

You should consider that most CSMA/CD networks although in theory using CSMA/CD in practice are really switched full duplex networks where collisions cannot occur. So, scaling up the speed does not change the situation at all.

What about real CSMA/CD networks, then? If you scale up the speed and keep the maximum allowed network size constant, you must scale up the minimum allowed packet size. Otherwise collisions cannot be detected. Eventually, the minimum allowed packet size exceeds whatever packet size is most common in your network, and then in that case you must insert dummy data to the end of the packet so that collisions can be detected. So, if increasing the speed is taken to the extreme, utilization will actually become lower.

1

There are several slightly different definitions of utlisation, for the purpose of this post I will assume it means the ratio of capacity used to successfully send useful data verses total low level capacity.

There is also the question of what the hosts are trying to do. If the hosts aren't actually trying to max the network out then obviously your utilitsation will drop as you increase the network speed regardless of low-level protocols.

The problem is you can't really scale up the speed of a CSMA/CD network while keeping "all other parameters unchanged". You must either increase the minimum packet size or reduce the maximum network size.

If you increase the minimum packet size then you end up sending more padding. Thus reducing utilisation.

If you reduce the maximum network size you don't have that problem but you may end up with a protocol that is unable to operate over a practically sized network.

This problem along with shifts in the relative costs of processing/buffering electronics verses line-driving electronics has caused CSMA/CD to fall out of favour. At 10M you could build a large CSMA/CD network. At 100M CSMA/CD was sometimes used but the maximum extent of a collision domain was far more limited meaning all but the smallest networks would have to use some bridges/switches. At 1G CSMA/CD was standardised but is pretty much never actually used. At higher speeds they didn't even bother to include it in the standards at all.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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