Comments on: Scoping http://jonathanlewis.wordpress.com/2007/01/08/questions-questions/ Just another Oracle weblog Wed, 19 Jun 2013 18:51:29 +0000 hourly 1 http://wordpress.com/ By: Jonathan Lewis http://jonathanlewis.wordpress.com/2007/01/08/questions-questions/#comment-1824 Wed, 17 Jan 2007 21:11:31 +0000 http://jonathanlewis.wordpress.com/2007/01/08/questions-questions/#comment-1824 Rob, yes the checks are OR’s. Conveniently you can read the source of the package – the specific condition (9.2) is:

and ( buffer_gets > l_buffer_gets_th
or disk_reads > l_disk_reads_th
or parse_calls > l_parse_calls_th
or executions > l_executions_th
or sharable_mem > l_sharable_mem_th
or version_count > l_version_count_th
);

]]> By: RobH http://jonathanlewis.wordpress.com/2007/01/08/questions-questions/#comment-1756 Tue, 16 Jan 2007 17:33:59 +0000 http://jonathanlewis.wordpress.com/2007/01/08/questions-questions/#comment-1756 Stupid question and I think I know the answer but:
“In the 9i version of statspack the resources checked, with their default collection boundaries”, these conditions are OR’s correct. As in as long as it meets one of the criteria.

Also, vidya, I’m often asked the same question and I’ve found it useful if snapshots are less than (or equal to) 15 min at most. I also only read in 15 min snapshots and check 15 min before and after the event (the after would support Jonathan’s post).

Many times if the snapshot is too long, single statements that are relatively large can fall down the report(or off) to the summation of something that runs and totals for multiple executions.

“The easiest way to find a needle in a haystack is to do a handful at a time.” – Grandpa (although he was referring to something entirely different that inefficient sql)

]]> By: vidya http://jonathanlewis.wordpress.com/2007/01/08/questions-questions/#comment-1464 Thu, 11 Jan 2007 20:34:11 +0000 http://jonathanlewis.wordpress.com/2007/01/08/questions-questions/#comment-1464 Jonathan, Thanks so much for explaining this – and sorry about being ambiguous – I did mean

“a single occurrence of a long running statement that is issued at 9:55, is still running at 10:00 and probably running past 10.15 am”

The reason I asked the question, frequently I am asked to look at performance for short windows

Questions like
“Vidya, can you see what happened between 6.00 and 6.30 am this morning the system was slow and users were complaining but things were fine after that?”

so I generate a statspack report for snap id’s between 6.00 am and 6.30 am – hoping that its going to give me all the information of what was happening at that time interval – I guess I will need to remember that sql calls can cross boundaries of a statspack report

Thanks again

]]> By: Ghassan Salem http://jonathanlewis.wordpress.com/2007/01/08/questions-questions/#comment-1346 Tue, 09 Jan 2007 14:47:08 +0000 http://jonathanlewis.wordpress.com/2007/01/08/questions-questions/#comment-1346 Recently, at a client’s site where snapshots where taken every 15 minutes, you could see that a certain SQL took X buffer gets when you print the statspack report from 9.00 to 12.00, and took Y>X buffer gets when you print the statspack report from 9.00 to 10.00. Looking into why I found out that the query was being flushed away regularly then rexecuted. This makes the figures for large time periods irrelevant.
This corroborates what you said in the statspack post about being careful when analyzing them

]]> By: Jonathan Lewis http://jonathanlewis.wordpress.com/2007/01/08/questions-questions/#comment-1335 Tue, 09 Jan 2007 12:46:27 +0000 http://jonathanlewis.wordpress.com/2007/01/08/questions-questions/#comment-1335 Ghassan, the joys of blogging: it ensures that someone’s there to cross the i’s and dot the t’s.

Yours was a point worth making – things can get flushed out of the library cache before they get into a snapshot that could report them. At least, as in scenario 4, there may still be some data about them in the underlying snapshot tables.

]]> By: Ghassan Salem http://jonathanlewis.wordpress.com/2007/01/08/questions-questions/#comment-1330 Tue, 09 Jan 2007 11:38:32 +0000 http://jonathanlewis.wordpress.com/2007/01/08/questions-questions/#comment-1330 Jonathan,
another case is when the query is no more in the sql area at 10.15, so it will not be in the snapshot, and thus will not appear in the report.

regards

]]>