[Server-devel] Serving 400+ students w/ a single central XS - ejabberd nightmare?
sverma at sfsu.edu
Tue Mar 10 00:51:43 EDT 2009
On Mon, Mar 9, 2009 at 7:04 PM, Bryan Berry <bryan at olenepal.org> wrote:
> On Tue, 2009-03-10 at 09:58 +1300, Martin Langhoff wrote:
>> On Mon, Mar 9, 2009 at 12:11 AM, Bryan Berry <bryan at olenepal.org> wrote:
>> > I am worried about the XO's and not the XS.
>> Now you're starting to see what I've seen :-/ I also worry about your
>> APs and networking infra -- to support 400 active users you'll want at
>> least 8 APs. In more realistic terms, you'll probably need 12,
>> assuming a reasonably balanced load.
> We will have roughly 8+ AP's. We have found that off-the-shelf AP's can
> handle around 60-70 users. But that doesn't still doesn't solve the
> problem of the XO's getting bogged down by tons of ejabberd chatter.
> DSD: do you have any ideas about this?
> We are looking at about 100-150 students per school and connecting 3-4
> schools to a central XS.
>> As I mentioned before... I am working on xs-0.6, with the
>> moodle-ejabberd magic.
> That's great, but our pilot starts in a month but that doesn't fit our
> timeline. I don't want to send out a completely new, untested XS into
> rural parts of Nepal.
> Do you have any other suggestions fo us?
What if you had a small footprint box (like a soekris or routerboard)
at the school that talks to APs on one end via a switch, and does
tunneling back to XS in a central location? That way you would have a
fairly dumb tunnel unit at school (literally plug-and-play) and XS
management back at your central shop.
Dr. Sameer Verma, Ph.D.
Associate Professor of Information Systems
San Francisco State University
San Francisco CA 94132 USA
>> > This dell server has a dual-core Xeon 3.0 GHz processor and 2 GB RAM.
>> > RAM was fine, beam only used 450 MB according to ps_mem.py
>> Well, that's 1/4 of your RAM. You need to budget for apache/php,
>> postgres, and squid. Right now the main problem is Squid.
> That was 450 MB during account creation. It dropped significantly
> thereafter. I didn't provide much server stats last time because the XS
> resource usage isn't a critical problem.
>> Thanks for the list below. Quite a few are about stuff I can't help
>> with (roof leaks, power cords...) the others, I'm working on...
> That's the whole point of why I am telling you about such problems and
> how they make a centralized XS easier for us to maintain.
> Bryan W. Berry
> Technology Director
> OLE Nepal, http://www.olenepal.org
> Server-devel mailing list
> Server-devel at lists.laptop.org
More information about the Server-devel