On Thursday 22 November 2001 05.21, Adrian Chadd wrote:
> I'm sure it'll go past 80 with the CPU pegged - just the service times
> will blow out a little. I'll run it on a more powerful polygraph
> machine tonight when I get home.
A 500 MHz celeron should be capable of quite a bit more than 80 before the
CPU becomes a bottleneck for service times.
Note: The way aufs is designed will waste some CPU when under load buzy
polling both I/O threads and filedescriptors, usually resulting in that you
can push the box a fair bit above where the CPU is fully utilized before
service times start to drop significantly.
When I last was measuring these things I was using a polymix-3 based boiler,
slowly increasing the request rate until there was errors. Requires you to
first fill the cache, and in each test run you need to wait for the WSS to
freeze before starting the boling phase. A quite effective tool in finding
out approximately where the limits are for
* Response times
* Hit rate
* System flatline
Regards
Henrik
Received on Thu Nov 22 2001 - 01:23:31 MST
This archive was generated by hypermail pre-2.1.9 : Tue Dec 09 2003 - 16:14:38 MST