I'm seeing a lot of confusion in this thread.
In the past, we have had real problems with wireless.
802.11 implementations are too easy to confuse by stations with random
settings, we have seen our share of stations that switched to
ad-hoc/IBSS mode when there were connection problems, drawing other
stations into their little disconnected bubbles as well.
I'm not sure we had this problem this time.
The base stations at IETF62 worked quite well, the signal was fine
almost everywhere, and more people had configured their systems not to
start ad-hoc networks (and not to join them).
I haven't even seen a rogue DHCP server this time.
To stamp out one urban legend in the making: 802.11b and 802.11g
interoperate really well (if configured correctly on the APs); I'm
don't think any of the effects I have seen can be explained by the
increased presence of .11g stations or the ability of the APs to use
higher rates with them.
.11a is welcome as there is lots more spectrum (and the decreased range
also helps to reduce interference). Again, no problem at all; there is
no way a .11a network can disturb a .11b/g one (unless the station
implementation is confused; in this case it should suffice to simply
switch off .11a on the station, or .11b/g for that matter).
So, applying "if it ain't broke, don't fix it": The wireless part of
the wireless network was fine, dont' try to "fix" it.
The main problem we had seemed to be one of too much intelligence in
the infrastructure connecting the APs.
A situation where ARP works without fail but IP packets don't get
through is not a problem with the wireless part.
It is sure nice to have an infrastructure that catches rogue DHCP
servers, localizes/caches ARP traffic etc.
It is even nicer, though, when the infrastructure just works.
Ietf mailing list