[Testing] first play with new XO 1.5 machines

[Ed McNierney]

On Oct 21, 2009, at 4:02 PM, Tiago Marques wrote:

> Hasn't mesh proved useful in deployments?

No, it hasn't.  We actively discourage anyone from using it, and I'm  
not aware of any successful applications of it except in test and  
demonstration situations.

I suspect you are, like many, many people, confusing mesh with ad-hoc  
networking and collaboration, as was mentioned earlier.  Mesh refers  
specifically to 802.11s networking, allowing XOs to act as relays  
forwarding packets from point A to point B where point A and point B  
are out of range of one another but can communication with other nodes  
in between.  There are serious algorithmic problems with using this  
technology in XO situations.  If you have 20 XOs in a classroom  
operating in mesh mode, they can collectively saturate the radio  
spectrum by all trying to "help" forward packets for each other.  It's  
been hard to design a model that allows nodes in a mesh to know when  
to STOP serving as mesh nodes; and if we were able to do that, it  
would be again difficult to figure out when they should turn  
themselves back on.

As far as I am aware, all XO networking scenarios currently in use in  
the field with the XO-1.0 can be replicated with XO-1.5.  We  
sacrificed the non-working mesh capability in favor of a different  
wireless radio from Marvell that saves us about 150 mW, and that kind  
of power saving is much more valuable to our users.

I also want to point out that in cases where 802.11s mesh operation is  
desirable, the Open802.11s stack (http://www.open80211s.org/) should  
be investigated.  There's no reason Open802.11s software can't be used  
to make an XO a Mesh Portal Point (MPP).  The only thing we're really  
losing in XO-1.5 is the ability of an XO to serve as a MPP in low- 
power mode, when the laptop is otherwise asleep.  It would be great to  
get that working someday, but not today.

	- Ed