[Olsr-users] preventing TC mess from packet loss ???
Tue Sep 2 04:20:35 CEST 2008
Bernd Petrovitsch wrote:
> Routing tables of two hosts may be inconsistent at any time. But that is
> not a unique "feature" of OLSR but IMHO of (almost?) all IP-routing
> protocols (including the "old ones" from the cable world like RIP and
> In the wireless world, it is just more visible because the latency and
> packet loss are (much) larger than on Gbit cables.
Exactly. I see strange things in routing table when OLSR working with
lossy links so I just thought that it could be good if there is a
solution to prevent TCs from losses.
> The basic problem is that you have rather large networks. As soon as you
> implement absolutely synchronous routing tables between an|y 2 hosts,
> you have one through the whole network.
> I have no idea how slow and sluggish any route change will be.
> IMHO you need *at least* some kind of "two-phase commit" (similar to
> databases) with all it's drawbacks. And you have to live the "two-phase
> commit" over slow and lossy links (and not only - like DBs - on locally
> 0% loss and quite fast disks).
> Actually I don't know if there are any papers on that subject ("reliable
> two-phase commit over slow, high-latency, lossy networks" or so) ....
More information about the Olsr-users