[Olsr-dev] Using OLSR and avoid paths with the same cost. Is possible ?

Markus Kittenberger (spam-protected)
Mon Mar 15 16:27:13 CET 2010


i think should work well

regards Markus

On Mon, Mar 15, 2010 at 4:02 PM, ZioPRoTo (Saverio Proto) <
(spam-protected)> wrote:

> Hello,
>
> I'm using OLSR for something not usual.
>
> I'm using OLSR to route traffic inside a VPN Overlay network. I'm
> implementing a theoretical work to write a scientific paper, I don't
> just need things to work, but I need to be exactly sure of what
> happens in the network.
>
> I have basically no packet loss on my VPN links, so all links are ETX
> = 1. I don't use the hysteresis because I know that code is not well
> tested. So I'm using the default configuration file shipped with
> olsr-0.5.6.r8, basically specifing just the network interface where
> olsr must run.
>
> Now when a generic node "A" wants to send traffic to node "B" there
> might be two different paths with extactly the same number of hops. If
> all links are ETX = 1 it is like hopcount.
> Now for my experiments to be rigorous I need to be extremely sure
> about what path OLSR will choose to route the traffic, even if there
> are some "same cost" paths.
>
> To avoid "same cost paths" my idea is to use the LinkQualityMult in
> the config to put a little "noise" in the links so that there are no
> links with exactly the same cost. Using:
>
> LinkQualityMult default RANDOM
>
> Where RANDOM is a number very close to 1 (between 0.95 and 1.05), and
> of couse this number is different in the configuration of every node,
> I should have unique cost paths ...
>
> Do you think this can work ?
> Anybody has a better suggestion to solve this problem ?
>
> thanks ! :)
>
> Saverio
>
> --
> Olsr-dev mailing list
> (spam-protected)
> http://lists.olsr.org/mailman/listinfo/olsr-dev
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.olsr.org/pipermail/olsr-dev/attachments/20100315/8cd40dcb/attachment.html>


More information about the Olsr-dev mailing list