job screening question

Owen DeLong owen at delong.com
Fri Jul 6 03:28:01 UTC 2012


On Jul 5, 2012, at 8:05 PM, William Herrin wrote:

> On Thu, Jul 5, 2012 at 10:25 PM, Owen DeLong <owen at delong.com> wrote:
>> On Jul 5, 2012, at 5:50 PM, Scott Weeks wrote:
>>> --- bill at herrin.us wrote:
>>> From: William Herrin <bill at herrin.us>
>>> 
>>>> 5.      What is the reason for the 100m distance limit within an ethernet collision domain?
>>> 
>>> What's an ethernet collision domain? Seriously, when was the last time
>>> you dealt with a half duplex ethernet?
>>> -----------------------------------------
>>> 
>>> 
>>> Now if someone answered it that way, I'd definitely be
>>> interested while the HR person would just hang up...
>> 
>> +1 -- That would be a perfectly valid answer and one of the list of answers I would actually give to HR.
> 
> Incidentally, 100m was the segment limit. IIRC the collision domain
> comprising the longest wire distance between any two hosts was larger,
> something around 200m for fast ethernet. Essentially, the collision
> signal caused by receiving the first bit of the overlapping packet had
> to get back to the sender before the sender finished the 64-byte
> minimum-size packet. Allow for the speed of light and variances in the
> electronics and that was the width of the collision domain.
> 
It was, but only if the device in between segments provided "retiming"
which basically meant collision-handling buffering.

The requirement was (IIRC) that the preamble traverse the entire wire
so that everyone could hear it and back off before data hit the wire.

Bonus points for knowing that a "late collision" describes "hearing" a
collision after you started transmitting data.

> Carrier sensing multiple access with collision detection. CSMA/CD. I
> haven't thought about that in a long time.

Heh... It still has its uses, even in human conversations. ;-)

Owen





More information about the NANOG mailing list