[K12OSN] Benchmarking k12ltsp

[Robert Arkiletian]

On 6/10/05, John Baillie <jbaillie at stmarys-school.org> wrote:
> The question of performance vs. price, scsi vs. sata, software raid vs.
> hardware raid, 100 Mb vs. 1000 Mb to the switch comes up on this list
> enough to bring me to the following.
> 
> Would it be reasonable / useful to come up with some sort of k12ltsp
> benchmark procedure?
> 
> I'm thinking of something where you boot up a given number of terminals,
> log on then run a script that runs a given set of tasks.

THis is a great idea! Then we could really compare apples to apples.
When posting the results we would need to include consistent system
specs.

k12ltsp version:
cpu type:
# of cpus:
ram:
fsb speed:
motherboard:
eth0 nic:
HD controller:
HD type:
# of hard drives:
HD configuration: ie raid 0/1/5/ separate /home drive
nfs mount of home: (yes/no)
switch:
switch uplink speed:
client nics:
client ram:
client cpu: (probably not needed)

One thing which cannot be used is external websites. If firefox is to
be used, and I think it should, then the site has to be downloaded and
saved locally first cause you will get different times based on net
traffic/ time of day. You want to keep the test totally internal. We
could also see how Centos compares to Fedora and eventually Ubuntu.

All we need is a good script to put a system through it's paces. Any
good script writers up for the job?

-- 
Robert Arkiletian
C++ GUI tutorial http://fltk.org/links.php?V19