On 2/18/15, 9:52 AM, "Juliusz Chroboczek" <j...@pps.univ-paris-diderot.fr> wrote:
>>> - Babel will avoid creating loops even when reconverging, but might >>> create blackholes; IS-IS will collapse during reconvergence; > >> I would say the IS-IS will create micro-loops and blackholes during >> reconvergence. I wouldn¹t say it would ³collapse². Note that these >>issues >> have been addressed with extensions in SP networks but I¹m not sure we >> should consider this for the homenet discussion. Also, I¹d point out >>that >> ISIS would converge extremely fast in smaller networks. > >On wired networks, that's certainly true -- IS-IS and OSPF don't need to >guarantee good behaviour during reconvergence, since the combination of >stable links with a reliable flooding protocol gives high probability >bounds on reconvergence time. > >Is there any experimental data that measures the behaviour of IS-IS's >reliable multicast subprotocol over the 802.11 MAC, in the presence of >marginal links? Recall that 802.11 multicast is unprotected by ARQ, and >that excessive distributed traffic may cause beaconing to fail, causing >stations to disassiociate. I’ve only heard the stories of these problems in IEEE 802.11 networks. One would hope that basic multicast problems at Layer-2 wouldn’t permeate up into the routing protocol… Having said that, I know of no standard extensions to ISIS to address this problem. I‘ve also heard there are proprietary fixes that haven’t made it into the IEEE standards but by no means claim to be an expert in this area. Thanks, Acee > >FWIW, I've had some unpleasant surprises with earlier versions of Babel >causing the 802.11 MAC to collapse for minutes (!) after a major network >disruption (rebooting the whole network simultaneously). Fixing that >particular issue involved making reconvergence less aggressive, which of >course harmed convergence speed on wired links (which I don't care >about -- anything converges fast enough on Ethernet). Granted, that was >in IBSS mode, infra mode probably behaves better. > >-- Juliusz _______________________________________________ homenet mailing list homenet@ietf.org https://www.ietf.org/mailman/listinfo/homenet
