[Server-devel] Ultra Mobile XO Server

s.boutayeb at free.fr s.boutayeb at free.fr
Mon Jun 23 11:12:41 EDT 2008


Selon Yama Ploskonka <yama at netoso.com>:

>
> > We could have 2 options:
> > a/ a dedicated XO based XS server, to be used during classtime, serving
> > eventually a few (< 40-50) XOs, for example in a class) as well as the
> teachers
> > monitoring XO.
>
> One thing that concerns me here is connectivity.  The class XOs would be
> expected to connect to this server, right?
Different options are possible, depending of the activities involved:
- mesh on: e.g. for collaborative activities (either classwide or schoolwide)
- mesh off: e.g. for individual evaluations/exams
- (semi-public) WAN on: e.g. for activities involving different classes located
in distant areas, including in different countries (see our "Windrose" project)
- (public) internet on: e.g. communication, information retrieval, publishing
- internet off: e.g. for classwide or schoolwide activities not intended to be
shared beyond the class/school

> Using the mesh?  Wiki and experience say the mesh cannot handle 40 XOs,
> probably 15 is tops.

We have here in France 'nil' experience and we will value your input and
comments concerning the feasibility and scaleability of each scenario.
>
> Now, if an access point would do the job...  Can a wireless router
> operate as an AP by itself? it probably can, but of course then it would
> merely establish a local network, which in my opinion is not such a
> loss, though full access to the internet tends to be considered
> essential by many.  I'd like a school-wide or even area WAN to engage
> kids in collaborating beyond walls in any case, even without full
> internet.  The mesh has very short reach.
>
> > b/ a multipurpose XO with an alternate boot options, to be used by the
> teachers:
> > - with the kids, during classtime, for monitoring purposes,
> > - outside of the class, in order to prepare/evaluate the class activities.
>
> This looks mostly as a software issue, which I think deserves some
> attention.  Clearly moderating blogs, etc, would be helped by some
> specific tools the teacher needs but are not needed by kids' XOs.
>
I think, Uruguay with the Ceibal project has a valuable experience with Blogs.

> In any case the XS plan was that it would not have its own screen etc,
> but would be managed remotely, probably from teacher XOs?
>
> > In case a, booting from an external storage may be an option, especially if
> > additional storage is needed. The installation of the XS on the internal
> disk is
> > another option, eventually with additional storage for the user data.
>
> Then the kids are depending on that teacher XO to be available to access
> that data, which as an idea is sort of transformational, since the
> current canonical vision I have heard around here is that the server
> should be available 24/7, and that obviously would seriously hinder
> flexibility for the teacher.
>
> Yes, we would save $500 a pop per server, but would we really?
>
The school will decide how much they will spend. In France, the equipment of the
primary school is financed at communal level. Generally speaking, it will be
certainly good to offer multiple options for the deployments.

> Personally I believe a server that does the server job, and feeds
> several APs is the way to go, even if the server is not up 24/7 (for
> example to save power) or even just feeds a LAN or WAN and not off the
> internet (to save on VSAT or other such expenses)
>
> Leaving the teachers to have flexibility with their XOs to use them as
> needed and take them home in the afternoons is important.  Of course it
> all depends if teachers actually use them for something.  It appears
> that many of the XOs handed to teachers in Uruguay got Bitfrosted over
> the December-February vacations, for they did not go near their servers
> often enough or some such cause.
>
> It's still an interesting idea, which at least in my case reinforces my
> interest in researching more about APs and region-wide networks.
>
> Yama
>

Yes, we have to reflect more, and to experiment various scenarios and use cases.

Bests

Samy




More information about the Server-devel mailing list