[Commotion-dev] Throughput advantage to 802.11s support

[L. Aaron Kaplan]

On Feb 9, 2013, at 1:30 AM, Ben West <ben at gowasabi.net> wrote:

> I agree with Aaron than for uses beyond meshes limited to a single channel, max 32 or max 50 nodes is still the same arbitrarily low limit.
> 
> I am curious, tho, which approach would be more effective in getting this limitation (and its implementations) amended: A) don't use 802.11s at all, or B) use 802.11s as-is and push for more adoption of 802.11s among your peers nevertheless, so that the number of people potentially demanding this amendment increases.

I would rephrase your question:
"In which situations does 802.11s or any other layer-2 mesh make sense"?

There are advantages to having a layer 2 mesh: 
 * zeroconf works out of the box
 * your windows shares will appear everywhere (is that good or bad? ;-) 
 * the whole thing is essentially one big LAN -> this reduces complexity

Of course, you have the disadvantage of potentially lower scalability.

Maybe the combination of layer 2 and 3 meshes can become really interesting...

> 
> As for meshing across multiple channels (or media, for that matter), I started a thread on the olsr-users list asking about good approaches to building efficient bridges for hybrid mesh topologies.  The trelay package recently added to OpenWRT looks especially useful for this, and I think that tool was also mentioned in recent Battlemesh announcements.
> 
Indeed!
I think we should test it there. trelay came out of a discussion Xavier, Felix, Henning and me at the last WBM in the Bus from the mountains back to Athens.
Felix then hacked the solution in 30 minutes in the bus ;)
It is somewhat a shame that we did not try it yet, it has so much potential.
Especially the combination DLEP + trelay.


a.