Are Collision Problems Data Link or Physical?

Obliged

Are Collision Problems a Data Link or a Physical layer issue on the OSI model?

Ciscopimpenator

Data Link--Layer 2

With Ethernet, collisions happen when two NICs try to transmit at the same time... Layer 2 deals with addressing and interfacing with the hardware, whihc involves determining when to transmit to avoid collisions.
Look at MAC and LLC sublayer.

I just did a look through with my CCNA book to make sure.


Good luck

iproute

A collision occurs when two frames collide. Then of course, when a collision is detected, the back-off algorithm runs and the sending systems will attempt to retransmit at different times, CSMA-CD.

Cucumber

Yeah, collisions exist, or not, due to the behaviour of the Layer2 protocol.

For instance, Token Ring, which is a DataLink protocol, avoids collisions by allowing only one device to transmit at a given time; only the one "having the token" can send a frame. Ethernet, also a DataLink protocol, behaves the other way around, all the devices could send a frame a the same time if they wanted to.

redgoblin

I think collisions actually happen at both layers. When you actually consider what 'makes' a collison, its actually the fact that the voltage representing one frame 'collides' with a voltage representing another frame. The overlapping of 2 voltage signals results in a 'super' sized illegal voltage which cannot be interpreted at the receiving ends.

So in summary I reckon it happens at both layers

Ciscopimpenator

I think it happens at both layers as well, but CCNA wants you to think layer 2.

802.3 Ethernet also happens at layer 1 and layer 2. Layer 1 defines the physical standards.

Cucumber

Well, collisions do happen at Layer1, at the cable when several signals sum to one another, but the origin of the collision per se is not a cabling or signaling problem.

If you use an Ethernet Hub you will have collisions in the network. Then, using the same cabling, the same voltage levels, the same signaling and the same connector pinout. but using an Ethernet switch instead of a hub, collisions will be gone.

So it is not a Layer1 issue.

jasonboche

I like a good CSMA-CD vs. CSMA-CA discussion.

My old networking essentials book made the following analogy which I enjoyed:

Consider a dinner table full of guests. One of them would like to speak.

For the person who would like to speak:

CSMA-CD (collision detect) would be as if the person just starts talking, not checking to see first if anyone else is talking (this would also be like one of my co-workers at work). Of course, if there is an interruption (a collision), each person stops talking and waits a random amount of time before they start talking again so as to avoid future interruptions (collisions).

CSMA-CA (collision avoidance) would be as if the person first clears his/her throat at the table, as if to announce to everyone he/she is going to start talking. Everyone finishes their conversation to allow the new person to talk without interrupting.

Or I have it backwards (I'm going from memory here)

Netstudent

I'm pretty sure she knows what a collision and csma/cd is...The actual collision does occur on the physical wire, however the csma/cd algorithm is layer 2. So if a question asks what layer does a collison occur, then I would say physical. If a question asks what layer is responsible for csma/cd I would say Layer2. If you Have a "collison problem" that is going to create issues on both layers.

Ciscopimpenator

Cucumber gave a good explanation.

Stick with Layer 2 for the CCNA test.

malweth

I think of CSMA/CD as a very low-level Layer 2 process... it's about as close to physical layer as you get. The only real protocol involved (aside from the 802.3 frame being sent) is the jamming signal sent during a collision. That leaves you with media access (vice media control ).

--

When actually working on something like this (perhaps a new technology to take the place of CSMA/CD in a new medium), there would likely be a group separate from both physical layer and switching. This middle group would handle the access of the physical layer by the framing process. It's possible that the physical or switching group would take on the task depending on the medium being used.