Re: Database Buffer Cache Hit Ratio

Hans wrote:

>Is this in response to my comment? If so, you misunderstood. I meant
>'don't tune unless there is a legit reason, such as user complaints.'

Yes, it was (I didn't specify that you had said it because of google reply strangenesses - sometimes it quotes the message being replied to, where you can't see the other replies, and other times it opens a box with no quoting amid a page of replies displayed. I sometimes remember not to attribute from memory :-).

I disagree with the premise that a database with no complaints is close to being tuned correctly and therefore shouldn't be tuned. It just plain flies in the face of my experience, as well as some questions posed in cdos and elsewhere. CTD only applies if tuning has been done.

Now, whether someone who thinks hit ratio tuning is a good thing should be doing it, well, that's another question. Certainly an argument might be that a production db would be better off not made worse. A counterargument could be such a person might help and might learn something, or even both. Or might display enough ignorance to be fired when things go south, which might be a good learning experience for all involved.

A lot can be read between the lines of pcambraf's posts, and a lot can be read into them. Asking if dirty buffers are waiting on the dbwr - well, Sybrand answered that, and probably more could be said on the subject if the OP would post basic information like in http://dbaoracle.net/readme-cdos.htm#subj12 .

pcambref: Note that it is more complicated. Oracle is fairly smart about when it actually tries to write.
http://asktom.oracle.com/pls/ask/f?p=4950:8:::::F4950_P8_DISPLAYID:5671284058977 has some good details.

jg

--
@home.com is bogus.
Richard R. Schrock said he became interested in chemistry when he was
given a chemistry set as an 8-year-old, and at first liked to "blow
things up."