How to figure distance between first and last host on a segment?

I'm using Skillsoft to prepare for CCENT and I had the following question, which I've never seen before. I'll type it up here, because I'm not sure how they got the answer.

Q: There are a number of collisions on a segment between host, what is the best way to troubleshoot?
A: Confirm the distance between the first and last host

Never heard of this before. Is this a physical thing?

thanks

Find more posts tagged with

Save $250 on 2025 certification boot camps from Infosec!

Book now with code EOY2025

Button

Comments

tex3030

I heard in a classroom environment that if the cable becomes too long you can get an error known as a "late collision." This is when HostA is expecting a response, doesn't get one and resends a message, and it then collides with the response from the other device.

If you do show interface fa0/1 one of the things it will show you is the number of bad transmissions it has made. One of the items within this show command is late collisions - this is from memory though.

I have never seen that counter go up, the longest run of copper I have used is about 360-370 feet, POE will not run this distance, but data has no issues.

nbeacham

The questions doesn't state specifically late collisions though so it's hard to say for sure. Assuming checking the duplex settings isn't one of the choices then I would agree with checking the cable length.

instant000

tex3030 wrote: »

I heard in a classroom environment that if the cable becomes too long you can get an error known as a "late collision." This is when HostA is expecting a response, doesn't get one and resends a message, and it then collides with the response from the other device.

If you do show interface fa0/1 one of the things it will show you is the number of bad transmissions it has made. One of the items within this show command is late collisions - this is from memory though.

I have never seen that counter go up, the longest run of copper I have used is about 360-370 feet, POE will not run this distance, but data has no issues.

As a soldier, I had field exercises, and I would get the opportunity to test the theoretical maximum lengths of UTP CAT-5, as well as COAX. (we'd want to connect things with as few runs as possible.) Based on those experiences, the book answers are about right.

My first experience with theoretical maximum length of Coax was December of 2000, during a field exercise at Fort Carson. We had the CAISSI (think router/switch/PBX in a box, basically) We could run phones and internet from the thing in the field. ... Anyway, PFC Instant000 had to run cables. I was a 25B (74B) back then, so I knew what I was doing, but I was working with some 92A (the supply tech) NCOs. After I showed them a few things, and knew how to actually run the cable testers and configure the devices by instinct, they had a lot more respect for me.

When we went to the dust bowl (NTC) in 2001 (We basically had back-to-back field exercises at Carson, I did at least three exercises there, and I was only there for 13 months), they stationed me back at the nicer part, as there was more intensive computer processing there.

So, it just goes to show, if you do your job well, there are some rewards to it, and this was at E-3/E-4. ... hope I'm a little better than that, ten years later!

JockVSJock

instant000 wrote: »

Anyway, PFC Instant000 had to run cables. I was a 25B (74B) back then, so I knew what I was doing, but I was working with some 92A (the supply tech) NCOs.

Go Army, HOOAH!

I'm currently 25N.

Slowhand

Since the question is asking about collisions, you're probably going to need to read up on carrier sense multiple access with collision detection (CSMA/CD), from back in the bad old days before switches were the norm and you'd have a whole lot of hosts connected to repeaters or hubs. There was a distance limitation, but it looks like the test might mean that the distance between hosts would impact if the carrier sense used to check if the line is free to send packets doesn't reach all the other hosts before your NIC starts sending and the "jam time" runs out. (I'm guessing, it's been a while since I looked at anything running half-duplex.)

Here are a couple of places to read up on CSMA/CD:

Wikipedia
Cisco's DocWiki (This has some specifics about the length of network segments between sending hosts.)