Take the 2-minute tour ×
Server Fault is a question and answer site for professional system and network administrators. It's 100% free, no registration required.

I'm learning about Ethernet and I'm wondering if with several stations within the same Ethernet collision domain like pictured below, is it possible that a frame sent by station A is fully received by station B before a collision is detected (due to C or some other station also transmitting) and a jam signal is sent? In other words, can a station receive the same frame multiple times due to retransmissions?

-----------------------
|   |                 |
A   B                 C
share|improve this question

closed as off-topic by Chopper3, pauska, Nathan C, Ward, TheCleaner Jul 24 '13 at 14:43

This question appears to be off-topic. The users who voted to close gave this specific reason:

  • "Questions about hardware or software used in a home setting are off-topic because they require answers that may not be practical for the business and support professionals here. You should try asking on Super User instead." – Chopper3, pauska, Nathan C, Ward, TheCleaner
If this question can be reworded to fit the rules in the help center, please edit the question.

2  
You can't have read our help before posting, read it before doing so again; serverfault.com/help/on-topic - this isn't a teaching site. –  Chopper3 Jul 24 '13 at 10:26
    
@PaulK: Such situation is impossible thanks to minimal frame length. answers.yahoo.com/question/index?qid=20070827093139AAhV7yK –  MBu Jul 24 '13 at 10:36
    
@Mbu, ah yes I thought so, thank you! –  PaulK Jul 24 '13 at 11:39

1 Answer 1

up vote 8 down vote accepted

Note that this is a historic issue without any significance to modern network infrastructures. Ethernet networks are typically built with switches / bridges which create a single Ethernet collision domain for each connected station and typically operate in full-duplex mode, without ever employing CSMA/CD. Ethernet standards newer than 1000BaseX have dropped CSMA/CD and half duplex operation entirely.

But concerning the core of your question: there are a number of factors defined in the Ethernet standards preventing this from ever happening:

  1. the minimal frame length for Ethernet, which is 64 Bytes for everything up to 100BaseX, resulting in at least 512 bit times of transfer and is extended by padding to 512 Bytes for 1000BaseX (carrier extension)
  2. the maximum span for an Ethernet collision domain (2500m for 10Base5 and 205m for 100BaseT)
  3. the minimum nominal velocity of propagation for the signal in Ethernet cabling
  4. the maximum number of repeaters between every two Ethernet stations (4 with 10BaseX, 2 with 100BaseX)
  5. the maximum latency induced by a single repeater

Combined, this would ensure that a frame's transmission is never more than half finished before the frame start has reached the farthest edge of the Ethernet collision domain. The other half of the transmission time is being reserved for the collision signal to propagate back through the network, so collision detection is guaranteed to work under all circumstances.

Further reading: Charles E. Spurgeon, Ethernet: The Definitive Guide; Vijay Moorthy, Gigabit Ethernet

share|improve this answer
    
very informative answer, thank you syneticon-dj! –  PaulK Jul 25 '13 at 9:19
    
@PaulK please mark it as an answer if it answers the question for you. –  the-wabbit Jul 25 '13 at 9:51

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