thin client or stand alone - which is better?

Gavin McCullagh gmccullagh at gmail.com
Wed Jan 30 09:11:50 GMT 2008


Hi,

On Tue, 29 Jan 2008, Jim Hutchinson wrote:

> Yes, we have one server connected to our network and all the clients
> connected to the sever via a switch. The server has gigbit but not the
> switches and maybe not our upstream network switches. I'll have to run
> this option by our network people and see if it's doable. Thanks for
> the info.

The upstream switches shouldn't matter as far as I know but doubtless your
network guys know your place better than I do :-)  What _should_ happen (in
theory) is the server talks to the switch at gigabit rate but each other
port can only work at 100Mbit.  So the sum of data between your 20 thin
clients can exceed 100Mbit, but no individual thin client can (but no
individual one should).

This is in theory and I do use this topology but I must admit I've never
done a measurement to verify it really works this way.

> I can't say I'm 100% certain it's network but I've watched the load on
> the server and it doesn't seem to be that. It happens whenever
> students are all using the internet. Could be firefox itself. I don't
> know how to test if we are maxing bandwidth but with one 100MB line
> coming in it seems likely.

If firefox seems responsive but is taking ages to load pages it might just
be that your upstream internet access is maxing out.  Firefox uses a lot of
RAM so you could be running out of RAM on the server.  I suspect a few
measurements would help.

Does flash get used a lot in the webpages (eg youtube, flash games)?  That
could be both a RAM hog and (if the stuff is animated) it could use a lot
of network bandwidth refreshing the screen.

> I'll check out the munin tool.

Always best to use real data to make upgrade decisions.

http://www.debian-administration.org/articles/229

Gavin




More information about the edubuntu-users mailing list