CSMA/CD 's definition based on wikipedia is :

Carrier Sense Multiple Access With Collision Detection (CSMA/CD) is a media access control method used most notably in local area networking using early Ethernet technology. It uses a carrier sensing scheme in which a transmitting data station detects other signals while transmitting a frame, and stops transmitting that frame, transmits a jam signal, and then waits for a random time interval before trying to resend the frame.

  1. I know about this algorithm. But what I am looking for is which devices on computer networking field use this algorithm to access the medium?
  2. Is switch use CSMA/CD algorithm to deal with collision on wired media ?
  • \$\begingroup\$ I believe this died with very early implementations of 10Gbit Ethernet, as it's a very inefficient and inconsistent algorithm to deal with collisions. Can you rephrase your second quesion? \$\endgroup\$ – user36129 Oct 20 '14 at 20:05
  • \$\begingroup\$ On my text books, I read about CSMA/CD, so they died for long time, what is alternative algorithm for them? \$\endgroup\$ – Mohammad Reza Rezwani Oct 20 '14 at 20:09
  • 1
    \$\begingroup\$ @user36129, I think you mean 2 Mb/s and 10 Mb/s ethernet. \$\endgroup\$ – The Photon Oct 20 '14 at 20:38
  • 1
    \$\begingroup\$ @alex, more recent ethernet isn't multidrop, just point-to-point, so CSMA/CD isn't needed. \$\endgroup\$ – The Photon Oct 20 '14 at 20:39
  • \$\begingroup\$ @ThePhoton: whoops... yeah, megabits, not gigabit :P Will Dean has his facts straight, that's exactly how I remember it. \$\endgroup\$ – user36129 Oct 21 '14 at 18:25

The answer to the first part of your question is not much nowadays, if we're talking about wired networks.

10Base-2 and 10Base-5 Ethernet were based on coaxial cable, with a shared medium (everybody talked and listened on the same wire), so used CSMA/CD.

10Base-T originally used 'hubs' which were just signal amplifiers/conditioners, and everybody on the network still used the same collision domain so needed the same rules, although at least the physical medium was separated and a fault on one connection didn't bring the whole network down.

Then the cost of switched Ethernet became low enough that hubs were pretty much universally replaced by switches - with a switch there is not one shared network or collision domain, but instead there's effectively one network per connected machine. Each of these little networks has just two devices on it - the connected computer and the switch port.

The switch then forwards packets from one mini-network to another, based on the destination address and an address table it builds-up in RAM.

So there is no concept of a collision at all, because there are only two devices on the network and they have a separate receive and transmit wires to talk on, so never interrupt each other.


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.