[Olsr-users] olsrd overhead
Wed Apr 8 20:31:15 CEST 2009
On Mittwoch 08 April 2009 20:24:57 Juliusz Chroboczek wrote:
> Ah, smart.
> So in short you flood a summary of the link-state database very
> frequently, and have some new kind of message to explicitly request
> reflooding of part of the database, right?
> Now, I'd expect one summary entry to take 6 bytes, which means that for
> a 100 node network, you'd be spending 600 bytes/s of traffic for
> distributing summaries. Which is definitely reasonable.
8 bytes, because we include a message type (to be able to use the same stuff
for other message types later).
> I'm wondering if you couldn't completely do away with periodic flooding of
> TCs -- TC flooding would only happen upon a topology change, or be
> triggered by explicit requests, when a node detects that there is some data
> that it doesn't have. Assuming aggressive duplicate suppression, this
> could be fairly efficient.
We plan to keep the compatiblity with the OLSR protocol, but maybe use huge
validity times (30 minutes ?) and reduce the number of periodic TCs to
something like 1 every 15 minutes (if nothing happens). Will need additional
work to run smooth, but I think that might be the next step to bigger
> Looks like a sexy protocol.
Yes, I think so. As long as the metric values of the protocol are stable
-------------- next part --------------
A non-text attachment was scrubbed...
Size: 198 bytes
Desc: This is a digitally signed message part.
More information about the Olsr-users