[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]

Re: [K12OSN] Tuning LTSP Performance

On Fri, 2008-08-29 at 06:16 -0400, Stephen Crampton wrote:
> I'm trying to get a thin-client lab up and running.
> Currently, I have a dual-processor AMD CPU with 4 GB of RAM as my
> server.  The server has a 1 GB Ethernet card.
> The server is connected to a 1 GB 8-port Ethernet switch.
> I have 7 clients attached.  They have Intel motherboards and are about
> 1 year old.  They have 512 MB of memory.  They have 100 Mbps Ethernet
> cards.
> Performance is sluggish, which surprises me.  When my students are
> running something very lightweight such as kbruch, the machines run
> fine, however, if one of them opens up TuxMath or TuxTyping or some
> other applications, everything slows way down.

Both of these apps are _VERY_ data throughput intensive. They send lots
of pixel data across the network. There are some tricks you can do.
1. Make sure the server is connected to a Gb port on the switch
(multiple ports is better using bonding). An 800x600 tuxmath uses
460,800,000 bits per second just for screen drawing at 60Hz refresh and
only 16 bit color (that's 56.25 MB/s!). I love tuxmath for it's teaching
ability, the fun math practice time and the kids love it. But it is a
> I noticed that there are parameters that can be tuned in lts.conf.  I
> reduced the screen resolution to 800x600 and the color depth to 8
> bits.  Does anyone have any suggestions for other parameters that
> might make my system a little more responsive?
> Also, I've been reading wiki pages and mailing list entries regarding
> fattening thin clients.  There seem to be two ways of doing it:  (1)
> building a new ltsp system "fati386" or (2) writing scripts that
> execute particular programs on the client and direct the display to
> the client's Xwindows server.
Tuxmath and tuxtype are prime candidates for local-app status. That
solves the network load issues nicely. Firefox and streaming multimedia
are local-app candidates for memory hogging and cpu hogging reasons.
> I've spent a number of hours on both approaches without success.
> Before I invest a lot more time, could someone recommend one approach
> or the other?
> Thanks,
> Steve
> -- 
> This message has been scanned for viruses and 
> dangerous content by MailScanner, and is 
> believed to be clean. 
> _______________________________________________
> K12OSN mailing list
> K12OSN redhat com
> https://www.redhat.com/mailman/listinfo/k12osn
> For more info see <http://www.k12os.org>
James P. Kinney III          
CEO & Director of Engineering 
Local Net Solutions,LLC                           

GPG ID: 829C6CA7 James P. Kinney III (M.S. Physics)
<jkinney localnetsolutions com>
Fingerprint = 3C9E 6366 54FC A3FE BA4D 0659 6190 ADC3 829C 6CA7

This message has been scanned for viruses and
dangerous content by MailScanner, and is
believed to be clean.

[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]