This problem was mentioned back in May (see
http://lists.wispa.org/pipermail/wireless/2006-May/025354.html ) but
it's still here, and I thought it might be worthwhile to bounce it off
folks again, to see if anyone has any new insights.
Occasionally, completely at random, many of our 2.4GHz APs start
high packet loss (usually 20-30%) and high latency (upwards of ten
seconds - not milliseconds, full seconds) for a few minutes at a time.
It always clears itself up eventually, usually after a minute or two,
but sometimes it won't go away for ten or fifteen minutes. It's
especially annoying when it happens in the middle of the day and all
those itchy business customers call in at once.
Random facts and tidbits:
* I've watched our network traffic, using Mikrotik's torch and StarOS'
beacon tools, and I don't see anything that looks like a DDOS, either
entering or leaving our network.
* It doesn't affect all our towers, just most of them. Specifically,
most remote towers (geographically, not in terms of network topology)
* (This is the Lonnie Nunweiler clause) It affects equally towers that
are bridged, routed, and hybrid bridged/routed.
* I don't think it's a backhaul problem, because towers on several
different kinds of backhaul links are affected simultaneously. (We've
got a mix of Alvarion and Trango gear for backhauls, and at least one
ancient YDI EX-1.) Also, there's no problem talking between tower
locations; pings between different APs take 10ms or so, just like they
* It only affects our 2.4GHz customers - folks on 900MHz gear, 5.3GHz,
and 5.8GHz connections don't have any issues while things are going
Basically, it looks like someone's got a big giant massive "something"
that spews out insane amounts of 2.4GHz interference, that somehow
knocks out all our customers within about a fifteen-mile radius, and
they turn it on every so often.
Does that conclusion sound reasonable? And if it does, what the heck
I do about it?