[MLB-WIRELESS] Cat5 max length

[paul van den bergen]

Hi joe,
just to clarify, the length in the standard says 100 m max per segment (from 
NIC to NIC, including patch panels, cables from wall to NIC, etc.) to 
maintain 100Mbps.  IIRC from my cisco course, assume 20 byte packet minimum 
or something. @ electron speed this can go some 150-200m @ 100 Mbps
20 (bytes) x 8 (bits/byte) x 1/100,000,000 (s/bit) x 1x10^8 (1/3 speed of 
light in m/s) = 160 m... so not far off.  note lots of assumptions here.
depending on the amount of contention you have to put up with, (i.e. if there 
are only 2 devices on the bus, not much) you can get up to 60-70% 
retransmittion... adjusting the MTU (I said MUA yesterday... too many TLAs) 
down helps.  signal loss is not really the problem...  anyhoo, if you are 
willing to put upa with a lower throughput, 10 Mbps for example (which has 
about 6Mbps effective traffic) then no hassle.  You should be able to go 
quite a looong way with cat5.  If you have access to a rj45 crimper, you 
should be able to make a cable run as long as you need and then if it does 
not work split it and insert a 10 Mbps switch in the middle...


On Fri, 24 Jan 2003 12:45 am, Joe Hovel wrote:
> Thank you all for the detailed information!
> Please correct me if I misunderstood anything. For my particular
> scenario:
>
> LAN with switch....long cable run....AP---Antenna on tall pole
>
> I learnt out of all that:
> 10Mbps may have a longer cable lengths then 100 (10 is OK by me, since
> my WAN link will only ever manage less than that).
> A (cheaper 10Mbps) switch (functioning as a repeater here) will manage
> the timing of packets between my LAN switch (existing 100Mbps) and the
> AP on the pole, the distance can then be more then 100m, because the
> response to any packet is acknowledged and verified by the (repeater)
> switch on each of the two legs from my (LAN) switch to the AP, before
> the next packet is scheduled to be sent. That way no collisions or
> unnoticed packet losses can occur.
> I understand that a hub can't manage that - if there were more than two
> ports in use, which is of course not the case here. Can someone try to
> explain why a hub cant manage two ports without collision or loss
> verification in time? I understood that a switch does this for many
> ports by remembering who's connected to all the ports and not bother
> sending packets to any other port than the one its intended for.
> In this case, it will be irrelevant how many ports packets will be sent
> to - only one (other than the sending one) will be connected....
> I'm not trying to be annoying about this or contradicting anyone, I'm
> trying to learn!
> Cheers,
> Joe
>
> > -----Original Message-----
> > From: owner-melbwireless at wireless.org.au
> > [mailto:owner-melbwireless at wireless.org.au] On Behalf Of Joe Hovel
> > Sent: Thursday, 23 January 2003 10:29 PM
> > To: melbwireless at wireless.org.au
> > Subject: [MLB-WIRELESS] Cat5 max length
> >
> >
> > Can someone remind me what the maximum length for cat5 is?
> > Also need to know if a switch or hub can act as a repeater to
> > extend the distance. Cheers, Joe
>
> To unsubscribe: send mail to majordomo at wireless.org.au
> with "unsubscribe melbwireless" in the body of the message

-- 
Dr Paul van den Bergen
Centre for Advanced Internet Architectures
caia.swin.edu.au
pvandenbergen at swin.edu.au
IM:bulwynkl2002
would somebody get this big walking carpet out of my way?

To unsubscribe: send mail to majordomo at wireless.org.au
with "unsubscribe melbwireless" in the body of the message