[olsr-dev] control packet overhead

[spoggle]

This is a complicated topic - there isn't any one right answer for it since 
the answer depends so much on the situation it's being used in. That's why 
(at least IMO) olsrd behavior is so configurable. Experimentation is crucial 
for optimal settings.

I use four scales when picking settings:

low overhead -> high overhead
stationary -> mobile
slow response -> fast response
stability -> instability

so higher mobility means faster response, and higher instability, and higher 
overhead. (Actually overhead is somewhat independent, I could choose slow 
response settings that used a large amount of pointless overhead, but not 
low overhead and fast response).

There are a couple of good papers at http://www.olsr.org/ - be sure to pick 
up the paper on the link quality stuff, it's good reading. These will help 
you understand better how this all works. Andreas and Thomas did a real good 
job on those docs.

dave c

On 6/8/05, Abhishek Misra <(spam-protected)> wrote:
> 
> 
> Hello,
> 
> I have just started working with olsr.
> 
> I was wondering what could be the acceptable performance from a routing
> protocol and quantify the same (e.g. acceptable control packet overhead or
> even cpu cycles).
> 
> Please throw some light on this.
> 
> Abhishek Misra
> _______________________________________________
> olsr-dev mailing list
> (spam-protected)
> https://www.olsr.org/mailman/listinfo/olsr-dev
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.olsr.org/pipermail/olsr-dev/attachments/20050608/84698e00/attachment.html>