Well that sounds just what were looking for!

Can you give an ideia of the troughtput of the Atheros 802.11a backhaul
links? Around here we have been using  Dlink 900AP+ doing wireless bridge
between nodes and we have about 1Mbyte/s on those point-to-point links in a
standard FTP transfer. The bigger links are about 4Km. We need to upgrade
those in order to achieve 3Mbytes/s, do you think its possible with Atheros
802.11a and m0n0wall?

Thank you,
Carlos Rosário


-----Original Message-----
From: Michael Mee [mailto:mm2001 at pobox dot com] 
Sent: sábado, 7 de Maio de 2005 8:20
To: m0n0wall at lists dot m0n0 dot ch; carlos dot rosario at netcabo dot pt
Subject: re: Wireless Access Point with M0n0wall 

On 5/4/05, Carlos Rosário <carlos dot rosario at netcabo dot pt> wrote:
 > Every board will have 2 CM9, one working in Access Point mode
 > and the other one in Client Mode.
 >
 > Will m0n0wall be appropriate for this?

We do this in our community wireless project (e.g. see 
http://socalfreenet.org/node/520 - both those boxes are running m0n0wall 
with two wireless cards: one CM9 for an 802.11a backhaul client, the 
other a Senao 2511MP+ (prism-based) as an AP.

And we're gradually converting other nodes to run this same setup (from 
www.nycwireless.org/pebble) for the sheer convenience of managing, 
tweaking etc.

Stability so far has been excellent. We've had no more problems with the 
1.2b7 builds than the earlier FreeBSD 4.7-based releases -- and we now 
do a lot more with them because of the Atheros 802.11a support (e.g. 
node-level bandwidth shaping, full snmp reporting, some PPTP configs).

The only fly in the ointment is the lurking 802.11b 'powersave mode' bug 
(http://m0n0.ch/wall/list/?action=show_msg&actionargs[]=113&actionargs[]=93)
. 
Whether this is a factor in the real world or not for our installations 
we don't know. Of the five busy APs we have running m0n0wall, one of the 
newest has locked up twice this month. But that could be a hardware 
problem. The others have been rock solid. So.... YMMV. Also, running 
dual CM9 cards might avoid this completely anyway - I *think* the 
powersave mode bug only applies to prism (and possibly Orinoco)-based 
chipsets.  (Jim, do you know?)

Hope this helped more than it confused!

cheers, michael
www.socalfreenet.org