[Server-devel] Reasonable number of users sharing an activity instance?

John Watlington wad at laptop.org
Wed Apr 6 03:57:09 EDT 2011


On Apr 6, 2011, at 2:22 AM, Sridhar Dhanapalan wrote:

> What is a reasonable number of users that one can expect to reliably
> share one activity?

It depends on wireless network bandwidth more than any other factor.

> Does an XS schoolserver make a difference in this
> regard?

Yes.

Using laptops registered to a school server and a single access point,
I was able to do simple collaboration between 70 laptops in a clean
RF environment.   Pushing it past this point tended to result in
laptops not being able to register with the school server, particularly
if they are all turned on around the same time.

See http://wiki.laptop.org/go/Collaboration_Network_Testbed for
some testing done with different network and server options a
while ago.

James' post provides many more details about why you need a
school server to reach these numbers, and other reasons why
you may hit limits at a lower number.

> Sugar doesn't seem to place any limits, so conceivably there could be
> 100 kids trying to access the same instance. I tried a game of
> Memorise on XOs in a class last week, and I saw errors such as
> children being able to connect to the session but not actually
> participate (which basically halted the game when it came to that
> child's turn).
> 
> This can be frustrating for the children and the teacher. If we could
> manage their expectations, they would be much more receptive.
> 
> Games in other media (board games, console games, etc.) cite a hard
> limit on the number of players (e.g. "2-4 players"). Might it be
> useful for us to do that in Sugar? This could be implemented as a mere
> recommendation or a hard restriction in the code.

Given that the problems tend to be bandwidth related, this might not help.
If not using a server, it is likely to make it worse, as each collaboration
will be multicasting.

Cheers,
wad



More information about the Server-devel mailing list