%wcache is too low

[Deepu Chakravarty]

Hi,
'sar -b 5 10' output giving %wcache =3. I have HP-UX 11.11 (rp5430) server with 5 GB RAM. I am running oracle9i(11i apps) on this server. dbc_pct_max is set to 10 and dbc_pct_min is set to 5(earlier 50 and 5 respectively). maxdsiz is 256MB. All the datas and indexes are distributed properly in different disks. I have 8 disks on the server.
%wio is always 0.
Why write hit ratio is too low ? As oracle prescribe '%wcache should optimally be between 65% and 85%'. Pls suggest how to overcome this problem.
Thanks in advance.

[RAC_1]

I do not think that this is a problem. I believe that oracle does it's own caching through SGA. The buffer cache setting is for OS. With wcache hit at 3 %, you can further decrease it and make some more memory available to users. Set it to 8 %.

Anil

[Massimo Bianchi]

Hi,
I agree with RAC.

How many updates have you got ? If you do not write much data, maybe that those performanceindicators are good enough.

How many log/switch per hour have you ?

Regards,
Massimo

[Deepu Chakravarty]

Hi,
log/switch is around 60 p.h.

[Massimo Bianchi]

Hi,
60 log/swichper hour is very much, you could have a log bottleneck.

Try to have just a log switch every 10/15 minutes. You may achieve this enlarging the redo. What is the size of your redos ?

FOr the cache: i also cache the document you are referring, and that are racomandation for oracle V7, rather old :).

IF %WIO is always 0, then you have no I/O bottleneck, and i would not touch anything.

HTH,
Massimo