Re: Performance Monitoring

On Oct 5, 4:37 pm, Jerome Vitalis <vitalismanN0S..._at_gmail.com.invalid> wrote:
> joel garry wrote:
> > On Oct 5, 4:08 am, Jerome Vitalis <vitalismanN0S..._at_gmail.com.invalid>
> >> Shouldn't point 2 read as follows?
> >> "[...] For example, if some silly SQL reads an entire table to get a
> >> few rows, you wil likely have a lot of unnecessary I/O that *will* fill
> >> up the SGA."
>
> > No.
>
> > For example, I ran this:
>
> > ...
>
> Ah, OK then! That was about FTS.
> In fact, I had completely misunderstood the purpose of your sentence. I
> thought you wanted to say that if some program happens to read many more
> rows than necessary, it would result in "useless" blocks entering the
> buffer cache.
>
> After re-reading your sentence, especially the "entire table" part, it
> now seems more obvious to me that you were referring to FTS. My bad.
>
> Thanks for your answer, Joel.

No problem, you could probably tell I was leading you on, since it is so easy to think Oracle does simple-minded things (which I inferred the OP was thinking). Astalavista's answer was exactly what the OP asked, and perfectly correct, but it implicitly confirmed misconceptions in my opinion.

There's also lots of stuff on metalink and OTN to see, though some of the older things on metalink may be somewhat misleading, and tend not to refer to third party techniques which may be better. Here's an interesting link:
https://metalink.oracle.com/metalink/plsql/f?p=130:10:3146610117385679612::::ALLTEXT,ANYTEXT,EXCTEXT,NONTEXT,KNOWLEDGE,PRODUCT_ID,PLATFORM_ID,SS,ST,NUMHITS,FILTCRIT,FILTCRITVAL:performance%20monitoring,,,,TRUE,,,0,DR,100,2,ST.Server.Performance.Database Also see the technical information under http://www.oracle.com/technology/deploy/performance/index.html I haven't tried Note:271064.1, but it sounds interesting.

Note that the "Separating Tables and Indexes" may be wrong or misleading in Note:148342.1. There may be correlated performance improvements simply by the distribution across devices, but the table and index separation is not why. There may be table and index accesses by different users at the same time, but that is a different issue. Also see http://sysdba.wordpress.com/2006/09/26/interview-with-tom-kyte/ .

jg

--
@home.com is bogus.
"I used to be a stoner until I saw Reverend Horton Heat" - Bumper
sticker