sumGirl wrote:

> One more thing (so far!) I noticed that while my cached read improved,
> my cached writes is down about 4-5%. Any thoughts on this guys? We do
> a lot of datamining and we are constantly loading flat files into our
> production database, so any kind of write is always a worry for me!
>
> <!> onstat -p
> .............
> Before:
> Profile
> dskreads pagreads bufreads %cached dskwrits pagwrits bufwrits %cached
> 900881808 1877905516 32921980794 97.26 65487771 101306873 1371660039
> 95.23
>
> After:
> Profile
> dskreads pagreads bufreads %cached dskwrits pagwrits bufwrits %cached
> 8432220 54223974 6035485 98.60 2390566 4134128 29294055 91.84

How long a time does it take to get to those numbers? How frequently
are you zeroing your statistics? On the data present, you have 600M
buffer reads and 30M buffer writes. That's a fairly typical ratio.
I've not had to worry too much about performance on production systems
for a while (so other people with current production experience can
contradict what I suggest if it is appropriate to do so), but with
20:1 ratio of reads to writes, I would not be too worried about the
write cache ratio. I would be most concerned about the read cache
ratio (percentage).

--
Jonathan Leffler #include <disclaimer.h>
Email: jleffler@earthlink.net, jleffler@us.ibm.com
Guardian of DBD::Informix v2003.04 -- http://dbi.perl.org/