[Olsr-users] Current OLSR protocol

[L. Aaron Kaplan]

On May 22, 2009, at 3:46 PM, Damian Philipp wrote:

> Hello,
>
> I'm currently gathering some information about available routing  
> protocols for mesh networks for my diploma thesis. So far, I have  
> found rather confusing things about OLSR. From http://www.olsr.org/?q=background 
>  I gathered that most of the optimizations to get olsr working were  
> code improvements. However, https://www.open-mesh.org/wiki/the-olsr-story 
>  paints a somewhat different picture with inherent instabilities in  
> the original concept of olsr which is why they started over with  
> BATMAN.


Well... Let me give you my personal opinion on that.
Mainly Elektra from BATMAN was very often behind a FUD strategy  
against OLSR.
Which is weird but it is like that. Maybe it is in order to start the  
new project BATMAN.
 From the OLSR.org side, this is great that there is a new attempt at  
routing protocols since we can all learn from that and from the new  
problems that a new routing protocol will give in practice.

However, I would *not* take the-olsr-story on open-mesh.org as  
correct, final, authoritative nor complete.
Fact is that mesh research is an ongoing process and we can all simply  
learn.

>
> From what I've read on the freifunk homepage, olsr is still widely  
> in use on freifunk networks.
it is.

> I'm guessing that they are probably using the latest stable olsrd  
> from olsr.org. However what I have not been able to find out is what  
> protocol the current implementation from olsr.org is using. Is it  
> still RCF-3626 compliant or does it rather contain all the  
> optimizations from the C-Base people, thus breaking RFC compliance?
>
it is RFC compliant but we all turn on extensions to RFC3626 in  
practice (ETX, Fisheye, ...) which break compatibility. But these are  
flags. You can turn them off.

> As part of my thesis I will have to experiment with some extensions  
> to the OLSR protocol. In order to be able to design those, I'd like  
> to know the protocol I'm actually using ;-)
>
  I suggest that you start to familiarize yourself with the basic  
datastructures, the RFC and maybe the plugin API. You will find more  
help concerning these issues on (spam-protected)

> I did find the OLSRv2-Draft at http://www.ietf.org/internet-drafts/draft-ietf-manet-olsrv2-08.txt 
> . This however still contains Multipoint relays, which as mentioned  
> on open-mesh.org seem to be (have been?) one of the bigger issues --  
> thus my confusion.

no, ... open-mesh.org is a lot of personal opinion by Elektra.
So, again... take that with a grain of salt please ;-)

There was even once a paper published by Elektra and some insitute in  
South Africa which again "proves" that OLSR does not scale and stuff  
like that.

In the history of routing protocols you can always find this trend  
that some proposals were made, some succeeded in acceptance (i.e. they  
were made an RFC) and then a *great* deal* of work was still needed to  
actually make it scalable and secure in practice.

Take BGP for example: we (== the internet) is still fighting with some  
basic security problems of BGP [1]. So even the development of BGP is  
still ongoing.

Summary: it is the implementation which counts!
And OLSR.org put *a lot* of effort into that the last 2 years.

So the basic real dichotomy between OLSR.org and BATMAN is actually  
the question:
  "do something new or improve the old stuff?"
I guess both are valid positions somehow.


[1] http://www.wired.com/threatlevel/2008/08/revealed-the-in/