On Thu, Sep 07, 2006 at 08:09:57AM +1000, ronnie sahlberg wrote:
> Since autonegotiate is unreliable :
>
> If you use autonegotiate there is a probability (small today) thet the
> peers will negotiate a duplex mismatch after each link up event.
Correct.
> This probability is small today but nonzero and depends on timings inside
> the chipsets, cable length, noice, bad luck etc
> Certain chipsets even today just can not negotiate properly and will even
> today get a near 1.0 probability of negotiating a mismatch.
>
>
> If you always have everything forced to full duplex there will never be
> any mismatches.
>
>
> Forcing everything to full duplex takes a small amount of extra work to do
> but makes the probability of a duplex mismatch zero.
While this statement is obviously true, it isn't easy (or cheap) to
achieve in some environments (e.g. university network with lots of
different PC/Workstation/Server/Thinclients/... admins - at least one
per faculty, sometimes one per Professor. Many will turn on autoneg even
if you instruct them not to do that, others will set 100F even if you
instruct them to set to auto. Since we actually started to set
everything to auto, the number of exceptions has decreased.
Ciao
Joerg
--
Joerg Mayer <jmayer@xxxxxxxxx>
We are stuck with technology when what we really want is just stuff that
works. Some say that should read Microsoft instead of technology.