From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from mail-dy0-f43.google.com (mail-dy0-f43.google.com [209.85.220.43]) (using TLSv1 with cipher RC4-SHA (128/128 bits)) (Client CN "smtp.gmail.com", Issuer "Google Internet Authority" (verified OK)) by huchra.bufferbloat.net (Postfix) with ESMTPS id 5C95E201A58 for ; Tue, 8 Nov 2011 11:39:27 -0800 (PST) Received: by dyf28 with SMTP id 28so78170dyf.16 for ; Tue, 08 Nov 2011 11:39:24 -0800 (PST) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=gmail.com; s=gamma; h=mime-version:in-reply-to:references:date:message-id:subject:from:to :cc:content-type; bh=ZPyas4AwzA623S6pA8XPYNA/grqOrHpX7At9BFeDBNg=; b=jmX3Zyr0jJBWiNCnURLDm4R5xXMPbkC13XRxXV+niTeNWOsGhA5s/Lpb2oggV9uWQV FZWN5YhH94o2d90cXeH/wnE0ztOJj2N7WjRmfGJidYDSUvAzmGupLKkf2ARKF5FyexBP WO7tCdWsORTZg4aWjfI0uzYPBe3wGSsEHv5Ic= MIME-Version: 1.0 Received: by 10.182.152.66 with SMTP id uw2mr10410488obb.17.1320781163865; Tue, 08 Nov 2011 11:39:23 -0800 (PST) Received: by 10.182.15.40 with HTTP; Tue, 8 Nov 2011 11:39:23 -0800 (PST) In-Reply-To: References: Date: Tue, 8 Nov 2011 20:39:23 +0100 Message-ID: From: Dave Taht To: Andrew Hammond Content-Type: multipart/alternative; boundary=f46d044401361924c704b13e546b Cc: bloat Subject: Re: [Bloat] small rural wireless ISP X-BeenThere: bloat@lists.bufferbloat.net X-Mailman-Version: 2.1.13 Precedence: list List-Id: General list for discussing Bufferbloat List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Tue, 08 Nov 2011 19:39:27 -0000 --f46d044401361924c704b13e546b Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: quoted-printable On Tue, Nov 8, 2011 at 8:05 PM, Andrew Hammond < andrew.george.hammond@gmail.com> wrote: > My friend owns and operates a small ISP that serves a small town and the > surrounding rural area. > > His current setup is wireless routers in bridge mode. > > Heh. I would suspect he has scaling problems. > As a first step, I'd like to move him to an OSPF routed network. I've > never built an OSPF network, can you guys please recommend a good page fo= r > reading up about it? > Cero's default routing protocol is babel, and is nearly zero setup. Change the ip addresses with a sed script, disable nat, plug them into each other. ospf and ospf6 are in the optional quagga package as are most other routing protocols. Years ago I started working with ospf but gave up on it due to requiring two different daemons, two different kinds of packets, and two different conf files for ipv4 and ipv6. It was too large to run both in 16MB of RAM, too. That latter problem is not a problem on larger routers. I later went to olsr and then batman and currently babel. There is seemingly big support for ospf and olsr within the theorists in the homenet working group. Babel, OLSR, and batman have respectable metrics for wireless networks. OSPF, not so much. Babel is 'RIP on speed'. I'm not going to go into babel's evolution (or religion) here, but I would like it if more people understood that being able to effectively route P2P links is crucial for wireless behavior.... http://www.pps.jussieu.fr/~jch/software/babel/ > > I'd like for the majority of the routers to be running cerowrt, however > that would require at a minimum external antenna connectors. > Ubiquity's products mostly use the same chipset, and have available external antennas or an all in one unit such as the nanostation M5. There are a lot of other products that use the same chipset, too. > I understand that the only hardware supported by cerowrt (so far) is the > Netgear WNDR3700v2. Is it (reasonably) possible to attach external antenn= as > to these routers? > The ideas of cerowrt are implementable in just about any hardware, but it's the ath9k chipset that matters at the moment, not the brand, and only because it's the most 'open source' of the wifi stacks we had to choose from when we started. I've periodically built a cerowrt version for the nano-m5, merely stripping out the named daemon support to gain enough flash for an 8MB system. It only takes a few minutes to make the change to the filesystem and package database and compile a version specific to that. Whenever cerowrt is 'baked' - which seems an increasingly distant goal of late - it will be easy to slam it on a couple dozen brands That said it remains our intent to push up everything possible into openwrt itself as it stablizes. I (or rather some enthusiasts did) pulled apart a 3800 and it showed the standard detachable tiny antenna connectors on the motherboard, and there are plenty of converters from that to a N connector - but ubiquity's gear already does that for you. > > Finally, I'm wondering how much interest there would be in some form of > access to this network for the purposes of analyzing or proving concept > code. > I certainly would like to have a long distance wireless network to play with again, and tools that monitored latency via things like snmp... I had had a mesh network using the nanostation M5s up in nicaragua 2 years back. got beat by bufferbloat. The ath9k series of drivers is fixed now 'enough' to go and deploy that again but I still seriously think active queue management is required now-adays to keep flows in hand. As another example I'd help build a gigE wireless backbone (wiline.com) around the bay area 6+ years back (has it been that long?). When I left the US they'd provisioned voice over the same backbone. When I came back, they'd had to go to a dual stack for voice support. Bufferbloat again... > > Sincerely > Andrew > > _______________________________________________ > Bloat mailing list > Bloat@lists.bufferbloat.net > https://lists.bufferbloat.net/listinfo/bloat > > --=20 Dave T=E4ht SKYPE: davetaht US Tel: 1-239-829-5608 FR Tel: 0638645374 http://www.bufferbloat.net --f46d044401361924c704b13e546b Content-Type: text/html; charset=ISO-8859-1 Content-Transfer-Encoding: quoted-printable

On Tue, Nov 8, 2011 at 8:05 PM, Andrew H= ammond <andrew.george.hammond@gmail.com> wrote:
My friend owns and operates a small ISP that serves a small town and the su= rrounding rural area.

His current setup is wireless rout= ers in bridge mode.
=A0

Heh. I would susp= ect he has scaling problems.
=A0
As a first step, I'd like to move him to an OSPF routed network. I'= ;ve never built an OSPF network, can you guys please recommend a good page = for reading up about it?

Cero's default routing protocol is babel, and is = nearly zero setup. Change the ip addresses with a sed script, disable nat, = plug them into each other.

ospf and ospf6 are in the optional quagg= a package as are most other routing protocols. Years ago I started working = with ospf but gave up on it due to requiring two different daemons, two dif= ferent kinds of packets, and two different conf files for ipv4 and ipv6. It= was too large to run both in 16MB of RAM, too. That latter problem is not = a problem on larger routers.

I later went to olsr and then batman and currently babel.

There = is seemingly big support for ospf and olsr within the theorists in the home= net working group.

Babel, OLSR, and batman have respectable metrics= for wireless networks. OSPF, not so much.

Babel is 'RIP on speed'. I'm not going to go into babel'= ;s evolution (or religion) here, but I would like it if more people underst= ood that being able to effectively route P2P links is crucial for wireless = behavior....

http://www.p= ps.jussieu.fr/~jch/software/babel/
=A0

I'd like for the majority of the routers to be runn= ing cerowrt, however that would require at a minimum external antenna conne= ctors.

Ubiquity's products mostly use the s= ame chipset, and have available external antennas or an all in one unit suc= h as the nanostation M5.

There are a lot of other products that use the same chipset, too.
= =A0
I understand that the only hardware supported by cerowrt (so far) is the N= etgear WNDR3700v2. Is it (reasonably) possible to attach external antennas = to these routers?

The ideas of cerowrt are implementable in just about = any hardware, but it's the ath9k chipset that matters at the moment, no= t the brand, and only because it's the most 'open source' of th= e wifi stacks we had to choose from when we started.

I've periodically built a cerowrt version for the nano-m5, merely s= tripping out the named daemon support to gain enough flash for an 8MB syste= m. It only takes a few minutes to make the change to the filesystem and pac= kage database and compile a version specific to that.

Whenever cerowrt is 'baked' - which seems an increasingly dista= nt goal of late - it will be easy to slam it on a couple dozen brands
<= br>That said it remains our intent to push up everything possible into open= wrt itself as it stablizes.

I (or rather some enthusiasts did) pulled apart a 3800 and it showed th= e standard detachable=A0 tiny antenna connectors on the motherboard, and th= ere are plenty of converters from that to a N connector - but ubiquity'= s gear already does that for you.
=A0

Finally, I'm wondering how much interest there woul= d be in some form of access to this network for the purposes of analyzing o= r proving concept code.

I certainly would like t= o have a long distance wireless network to play with again, and tools that = monitored latency via things like snmp...

I had had a mesh network using the nanostation M5s up in nicaragua 2 ye= ars back. got beat by bufferbloat. The ath9k series of drivers is fixed now= 'enough' to go and deploy that again but I still seriously think a= ctive queue management is required now-adays to keep flows in hand.

As another example I'd help build a gigE wireless backbone (wiline.com) around the bay area 6+ years back (h= as it been that long?). When I left the US they'd provisioned voice ove= r the same backbone. When I came back, they'd had to go to a dual stack= for voice support. Bufferbloat again...
=A0

Sincerely
Andrew

_______________________________________________
Bloat mailing list
Bloat@lists.bufferbloat.net<= /a>
= https://lists.bufferbloat.net/listinfo/bloat




--
Dave T=E4ht
SKYP= E: davetaht
US Tel: 1-239-829-5608
FR Tel: 0638645374
http://www.bufferbloat.net=
--f46d044401361924c704b13e546b--