FreeBSD 6 is coming too fast

[Kris Kennaway]

On Sun, Apr 24, 2005 at 09:35:20PM -0400, Mike Tancsa wrote:
> At 09:02 PM 24/04/2005, Kris Kennaway wrote:
> >>
> >> I also tried a CURRENT snapshot then and there wasn't much of a 
> >difference
> >> between it and RELENG_5.
> >
> >disk I/O or filesystem I/O?  It would be interesting to benchmark the
> >latter since all the recent VFS work.
> 
> This was on hardware using the 3ware driver which is essentially the same 
> on RELENG_4 and RELENG_5.

This driver was recently updated, BTW.

> I also tested IDE performance which gave similar 
> results (i.e. RELENG_4 and DragonFly was better), but the drivers are 
> different so its hard to gage if thats a driver issue or not.   I am not 
> sure if any of those tests answers your question, as I am not sure how to 
> answer it.  I was looking for a way to measure overall throughput that 
> samba, NFS, database and imap servers could do either on RELENG_4 or 
> RELENG_5 as we start to migrate various servers from RELENG_4 to RELENG_5.

Measuring disk device performance (i.e. running a benchmark against
the bare device) and filesystem performance (writing to a filesystem
on the device) are very different things.

> I have a faster disk subsystem I can test against (Areca SATA RAID) that 
> works on RELENG_4,RELENG_5 and HEAD and could re-run the tests varying just 
> the base OS.  If there is a particular test you feel best simulates disk 
> performance, I am happy to test.

As is well-known, doing meaningful (disk) benchmarks is hard, and it's
easy to draw incorrect conclusions if you don't understand exactly
what it is you're measuring.  I'm not an expert in that, but it's been
discussed on the lists before.

Kris
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 187 bytes
Desc: not available
Url : http://lists.freebsd.org/pipermail/freebsd-current/attachments/20050424/fa2e036b/attachment.bin