[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: OT question sorry but i need salution fast ....

Hall Stevenson wrote:
> > I'm not saying that this is an ideal solution, but could
> > you install a cheap 4-port switch (not hub!) every 100
> > meters?
> >
> > Anyone care to tell me if or why this is a bad idea?
> A device is a device, right ?? Whether it's a network card,
> hub, or switch, keep the distance between them under 100
> meters. Sounds logical ;-)

Wrong... The following:


is NOT the same as:


The problem is, a hub just electrically connects the signal (handling
crossover, etc). A Switch will completely regenerate the signal on the
other port(s). The overall length applies to each segment. The top
diagram is one segment, which can have an *overall* length of 100
meters, not 400 meters as your analysis might imply. The bottom diagram
can have 100m between the first PC and the first switch, 100m between
the two switches, and 100m between the 2nd switch and the 2nd PC
(assuming 10 base-T Ethernet).

The issue, IIRC, has to do with the minimum size of an Ethernet frame,
the speed at which the frame propagates down the physical wire, and the
need for all devices to be properly able to sense a collision. Basicly,
if the overall length of the network is too long, PC#1 could begin
transmitting a frame while listening for a collision. PC#2, if far
enough away, could start transmitting a frame at nearly the same time,
also listening for a collision. If the network is long enough to
introduce enough delay that the start of the frame transmitted by PC#1
did not reach PC#2 before PC#2 was done transmitting (and if the frame
from PC#2 did not reach PC#1 before PC#1 was done transmitting) neither
side would detect a collision. The result would be a corrupted packet,
which would hopefully be handled at a higher layer.

I'm a test engineer by trade, not a communications engineer, so my
explanation may be a bit off. Anyone have a copy of IEEE 802.3 laying
around? I assume that deals with this problem in more detail.


Rich Puhek               
ETN Systems Inc.         

To UNSUBSCRIBE, email to debian-user-request@lists.debian.org 
with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org

Reply to: