[OLPC Networking] Path discovery sanity

[Aaron Kaplan]

Ah!

I forgot...
What I really wanted to say - some experience from the OLSR-NG project -
most people talk about scalability of this and that protocol etc....  
However...
actually the far bigger effect and difference that I saw between  
different protocols is
a) the quality of the implementation and algorithm choice and b) how  
you chose your metric and c) how careful you are with your  
"airtime" (i.e. how much of the precious spectrum you conserve  
without flooding it (*)).

So: in other words: the proto itself matters not so much anymore.  
Above points matter.

:)

my 2 cents,
a.


(*)
That is why I for example started to believe that the b.a.t.m.a.n  
protocol does something severely wrong: they flood the whole net.
Every node repeats every hello message for every other node.  Yuck!  
There is something wrong with that.


On May 13, 2008, at 9:12 PM, Javier Cardona wrote:

> Scott,
>
> On Tue, May 13, 2008 at 10:23 AM, C. Scott Ananian  
> <cscott at cscott.net> wrote:
>>  It may also be worth comparing to a pure proactive approach
>>  such as OLSR, which is also permitted under 802.11s and may scale
>>  further than HWMP.
>
> The draft defines an "extensibility framework" for path selection
> protocols, which allows the definition of vendor proprietary
> protocols.  But the only path selection protocol proposed for
> standarization is HWMP.
> OLSR was included in the past as an optional alternative to the
> mandatory HWMP, but it was removed last year.
> The group concluded that there were no usage scenarios in which OLSR
> would provide clear benefits over (the proactive part of) HWMP.
>
>> In contrast, the author of the HWMP protocol only expects it to  
>> scale to 50 nodes (see attached).
>
> The original HWMP specification (
> https://mentor.ieee.org/802.11/file/06/11-06-1778-01-000s-hwmp- 
> specification.doc
> ) does not specify a limit in the number of nodes.  Bahr takes the 32
> node
> limit from the 11s PAR
> (http://standards.ieee.org/board/nes/projects/802-11s.pdf) which is
> actually
> a design guideline and not a theoretical limit.  In fact, that limit
> probably comes from the medium
> access mechanism and not from the path selection protocol.
>
>>  A reactive approach is much more sensitive to the loss of a single
>>  packet during route-finding;
>
> That's just because the proactive approach transmits more redundant
> information.
> The reactive approach can be made as "insensitive" by increasing  
> redundancy.
> The main trade-off between proactive and reactive methods is  
> latency vs.
> management overhead.
>
>>  Hybrid approaches (such as the one specified in
>>  802.11s but not actually implemented by OLPC) may combine the
>>  strengths of both.
>
> Yes.  Or the weaknesses.  It all depends how you tune it ;)
>
> Cheers,
>
> Javier
>
> -- 
> Javier Cardona
> cozybit Inc.
> _______________________________________________
> Networking mailing list
> Networking at lists.laptop.org
> http://lists.laptop.org/listinfo/networking
>

---
there's no place like 127.0.0.1