From jasper at pointless.net Wed May 1 00:11:44 2002 From: jasper at pointless.net (Jasper Wallace) Date: Sun Jan 19 18:24:11 2003 Subject: [consume-routing] Current status of mesh routing on consume? In-Reply-To: <20020425093816.GA6083@aldigital.co.uk> Message-ID: -----BEGIN PGP SIGNED MESSAGE----- On Thu, 25 Apr 2002, Kim Hawtin wrote: > > > At the moment there are no working, widely available routing protocols that > > > are sutible for consume, mainly beacause no protocol has sufficent > > > authentication to work in a network where the routers are untrusted. > > > > I think you're fundamentally screwed if you have to rely on nodes to be > > trusted (by humans, nodes themselves building trust between each others is > > fine). The default assumption is hostile, and the protocol has to be able > > to deal with disruptions intrinsically. > > > > DoS issues are best addressed with agoric load levelling, but I hear > > current protocols are easily disruptible at several points. Can you sketch > > out the classes of attacks on current routing protocols? > > I we keep having 'positive' feedback like this, then consume.net > will never get anywhere ... > > Why is it that noone trusts anybody in this miserable town? We do, but it's best to build systems that avoid being disrupted by what ever can be thrown at it. A better way of putting this issue might be: Routers in Consume should work something like an ant heap or bees nest - many individuals work together to acheive some common goal (getting packets where they are meant to go), and those individuals are capable of detecting activity that may interfear with acheiveing that goal and act appropiatly. - -- Information wants to be a lumberjack. [http://pointless.net/] [0x2ECA0975] -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.0.6 (NetBSD) Comment: For info see http://www.gnupg.org iQEVAwUBPM8WpQCB+Qwuygl1AQEtugf/U2ly3s023rVyJS96JJMmqfYeHFhnrYWJ VE9BZXB2FtZAFQcfS3avGfvLATGeXZcOzF+VX+rRyXROF3Fnli1zcspXRjmhJWDf wgSfs9s9JL8vH+C5QaJ2Wv4MblqjdjjI/GvVYDbTc++PNuTTgvLU+w0sqEz7zEI/ PQlZjImjfnnQQ+Z4NnQrZ90JG9aET170j8CBUemhnJdXPbOfB1XUc2q/vzhGKuZX baOYK08ezdIp4kTKzFV9N2HhHgMspUei+iH4HS7tCih744GjZacxC5tyIEJJVsX6 2CGlmlTTN9OkEhzsnyRrLIHTnmYcHFdpPyNvKWZH+PJrFcwZQyWX2w== =YwrR -----END PGP SIGNATURE----- From leo at bind.org Wed May 1 12:00:01 2002 From: leo at bind.org (leo vegoda) Date: Sun Jan 19 18:24:11 2003 Subject: [consume-routing] Discussion of 'Anderson mesh' ideas.. (was: mobile mesh under GPL) In-Reply-To: <20020430125546.GA65272@bind.org> References: <003101c1eb8c$2f4a6cc0$4f6464c0@semper> <20020430125546.GA65272@bind.org> Message-ID: <20020501100000.GB85945@bind.org> Hello, On Tue, Apr 30, 2002 at 01:55:46PM +0100, I wrote: Re: Re: [consume-routing] Discussion of 'Anderson mesh' ideas.. (was: mobile mesh under GPL) [...] > The new draft IPv6 policy can be read from mailing-lists/global-v6/archive/2002/04/msg00050.html>. It's been > approved in the ARIN and APNIC regions and should be discussed tomorrow > at RIPE 42 in Amsterdam. The policy was accepted by LIR Working Group. Regards, -- leo vegoda "One size never fits all" RFC1925 - The Twelve Networking Truths From kim at aldigital.co.uk Wed May 8 13:03:40 2002 From: kim at aldigital.co.uk (Kim Hawtin) Date: Sun Jan 19 18:24:11 2003 Subject: 'Anderson Mesh' geodetic? - (subject was:) [consume-routing ] RE: New Not "mobile mesh" released under GPL In-Reply-To: References: <002c01c1eb82$5a371640$4f6464c0@semper> Message-ID: <20020508110340.GB20075@aldigital.co.uk> On Wed, Apr 24, 2002 at 06:02:54PM -0400, R.D. Hammond wrote: > 10.28.34.21 has gotta stamp the routing packet 'via 10.28.34.21' > to make sure, that 10.28.34.21 doesnt get that route back from a > different neighbour and asume it is a new route ... > > circular routes without end. Is it possible to set the hop count to a sensible number and decrment it, when it hits zero, drop the packet? Maybe I should refrase that, is the right thing to do the decrement the hop count? The problem is that it would have to be a high number if the mesh is large. Especially if there are no tunnels... yours, Kim -- :Kim_Hawtin:--------------------------------------:-----------------: | A.L. Digital Ltd. Tel: +44 (20) 8742 0755 | .^. Don't fear| | The Stores Fax: +44 (20) 8742 5995 | /V\ the | | 2 Bath Road http://www.thebunker.net | // \\ penguins!| | London W4 1LT http://www.aldigital.co.uk | /( )\ | | UNITED KINGDOM mailto:kim@aldigital.co.uk | ^^ ^^ | From jon at locust.co.uk Wed May 8 17:19:31 2002 From: jon at locust.co.uk (Jon Anderson) Date: Sun Jan 19 18:24:11 2003 Subject: 'Anderson Mesh' geodetic? - (subject was:) [consume-routing ] RE: New Not "mobile mesh" released under GPL In-Reply-To: <20020508110340.GB20075@aldigital.co.uk> Message-ID: <006201c1f6a3$c0f61760$4f6464c0@semper> It isn't quite that simple.... The crux is: How you know when to accept an updated route? I have one method using distance vector, which looks at the ICMP traffic and has a confirmation when new routes are to be accepted. This would work, but is a bit of a hack and involves the exchange of possibly unnecessary data. My current solution involves probabilistic routing tables and geodetic aware co-operative mobile agents to build routes. I'm also working on a full simulation for the scaling tests. Best Regards, Jon Anderson http://Locust.Net > -----Original Message----- > From: Kim Hawtin [mailto:kim@aldigital.co.uk] > Sent: Wednesday, May 08, 2002 12:04 PM > To: R.D. Hammond > Cc: Jon Anderson; 'Eugen Leitl'; > consume-routing@lists.consume.net; MOBILEMESH-LIST@lists.mitre.org > Subject: Re: 'Anderson Mesh' geodetic? - (subject was:) > [consume-routing ] RE: New Not "mobile mesh" released under GPL > > > On Wed, Apr 24, 2002 at 06:02:54PM -0400, R.D. Hammond wrote: > > 10.28.34.21 has gotta stamp the routing packet 'via 10.28.34.21' to > > make sure, that 10.28.34.21 doesnt get that route back from a > > different neighbour and asume it is a new route ... > > > > circular routes without end. > > Is it possible to set the hop count to a sensible number and > decrment it, when it hits zero, drop the packet? > > Maybe I should refrase that, is the right thing to do the > decrement the hop count? The problem is that it would have to > be a high number if the mesh is large. Especially if there > are no tunnels... > > yours, > > Kim > -- > :Kim_Hawtin:--------------------------------------:-----------------: > | A.L. Digital Ltd. Tel: +44 (20) 8742 0755 | .^. Don't fear| > | The Stores Fax: +44 (20) 8742 5995 | /V\ the | > | 2 Bath Road http://www.thebunker.net | // \\ penguins!| > | London W4 1LT http://www.aldigital.co.uk | /( )\ | > | UNITED KINGDOM mailto:kim@aldigital.co.uk | ^^ ^^ | > >