From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from mail.lang.hm (unknown [66.167.227.145]) (using TLSv1.2 with cipher ADH-AES256-GCM-SHA384 (256/256 bits)) (No client certificate requested) by lists.bufferbloat.net (Postfix) with ESMTPS id 0DAAF3B29E for ; Mon, 10 Jul 2023 23:02:40 -0400 (EDT) Received: from dlang-mobile (unknown [10.2.2.69]) by mail.lang.hm (Postfix) with ESMTP id D566B19C926; Mon, 10 Jul 2023 20:02:38 -0700 (PDT) Date: Mon, 10 Jul 2023 20:02:38 -0700 (PDT) From: David Lang To: Bob McMahon cc: David Lang , Aaron Wood , Make-Wifi-fast In-Reply-To: Message-ID: References: <29385648-o20s-nn8q-2p3s-900sn8606857@ynat.uz> MIME-Version: 1.0 Content-Type: text/plain; format=flowed; charset=US-ASCII Subject: Re: [Make-wifi-fast] I used to dream of a single wifi cpu, memory, and I/O X-BeenThere: make-wifi-fast@lists.bufferbloat.net X-Mailman-Version: 2.1.20 Precedence: list List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Tue, 11 Jul 2023 03:02:40 -0000 I'm providing the equipment (still 3800s running openwrt). I did a presentation at LISA in 2012 and a paper in ;login going over a lot of this. if you don't get the RF side right, you have no chance of having a good network under heavy load. David Lang On Mon, 10 Jul 2023, Bob McMahon wrote: > That you can build these networks to operate well says more about you than > the guys supplying parts & equipment. Impressive real world knowledge for > sure. > > Bob > > On Mon, Jul 10, 2023, 4:34 PM David Lang wrote: > >> First off, I am a Huge proponent of getting the RF layout clean before >> anything >> else. Directional signals sound like a fantastic idea, until you realize >> that >> the stations you are talking to are not using directional antennas, then >> the >> value drops significantly (under these conditions, directional antennas >> create >> more hidden transmitters) >> >> When I setup conference wifi, I take advantage of the fact that some walls >> block >> the wifi signals, so I can put the APs closer to the walls that block them >> without worrying about what's on the other side. I also place them on the >> ground >> (under the chairs) as the bodies of the attendees absorb the signal and >> let me >> use more APs in a room than I could get away with otherwise. I also stick >> to the >> standard 10MHz channels, as that lets me re-use a channel with more >> separation >> between APs using the same channel >> >> And yes, I agree that it makes no sense to run an AP at a higher power >> level >> than the stations talking to it, so I turn the power way down. >> >> I haven't knowingly run into the problem you describe of clients >> enumerating all >> available APs, but I may not have setup a high enough density of APs to >> run into >> the problem, can you give more info on that? >> >>> There is no reason to send energy more than 29' as that's the distance >> per >>> fire code that a human has to be from a working smoke detector. and in >> many >>> cities, one can't sell a house without a hard-wired, battery backed up, >> and >>> inter connected smoke detectors. >> >> As someone who has been looking at building a house, it's not that simple. >> Hard >> wired smoke detectors are only needed in some rooms, not in all, and only >> if the >> house is above a minimum size. Then you need different amounts of power to >> get >> through walls depending on how they are built. >> >> When I setup wifi in a conference center exibit hall that's 25,000 sq ft, >> I >> don't believe that there are 80 smoke detectors in that one room (not to >> mention >> the fact that the cealing is more than 29' away, even if I'm standing >> directly >> under it) >> >> Then there's the fact that not everything is inside. >> >> David Lang >> > >