A recent thread on the Oracle newsgroup comp.databases.oracle.server started with the following tkprof summary from a trace file: (more…)
June 28, 2007
March 14, 2007
Or as Doug Burns put it on his blog: how can I tell the actual DOP used for my parallel query”
As Doug points out, you first have to ask “When are you asking the question”. Immediately after you have run the query you could check v$pq_tqstat to see what happened, but apart from that you have very little hope of finding out in retrospect what happened in a specific query.
February 18, 2007
A few days ago, I described a change to Statspack in 10g that allowed you to see how much time each snapshot took. In that article I suggested copying the code back to 9i; however this is a fiddly little task, and you may prefer to avoid it.
Here’s an alternative – actually a strategy with a much wider use – which does the same, but gives you even better information. Use a “before logoff” trigger to record the information you need. There are lots of ways you could create a suitable trigger; here’s an example – with just a couple of naughty features that you might want to change:
February 12, 2007
Following yesterday’s note on SQL Profiles, someone asked how I detect that an opt_estimate hint had been used – with specific reference to the “index_scan” option. The reason for this particular choice is that other opt_estimate hints have a highly visible impact in the 10053 trace files, but this one doesn’t.
January 18, 2007
A little while ago I wrote a note about an anomaly that someone was seeing with the index_ss() hint. Since then, they forwarded me a 10053 trace file of the question.
I’ve posted brief analysis of it on my website, just to give you an idea of the sorts of things you can look out for when you’re trying to work out why Oracle seems to be ignoring a hint.
January 15, 2007
January 12, 2007
Just a little follow-up on my earlier note on buffer sorts. The following is an extract from a a tkprof output showing the rowsource operation for a query. Note especially the rows information. (more…)
January 11, 2007
If you look at a 10053 trace file, you will find references to Resc and Resp (sometimes rsc and rsp) lurking in the background whenever the trace file mentions Cost. When you get to the 10gR2 trace file, you see these described in the “Legend” section as: (more…)
January 5, 2007
I have made a few comments in previous articles about the use of bind variables and some of the peripheral details that can introduce surprises; and in the article on superfluous updates I made a throwaway comment about getting multiple child cursors for a single statement if you had columns of varchar2() or nvarchar2() defined to be longer than 32 bytes. It’s worth expanding on this point.
December 17, 2006
In an earlier article I mentioned the buffer sort in a footnote; I thought I would expand a little more on what I think it does and why it appears as a buffer sort in an execution plan rather than the more traditional sort (join).
Consider the trivial script:
December 11, 2006
I’ve just received an email from someone who had been on my “masterclass”, and he had a question about one of the comments I had made, namely: “bind variable peeking is always done, even if histograms are not generated”.
November 27, 2006
From 9i onwards, if you enable event 10132 in your session, then every statement you subsequently optimise will be dumped into your trace file, along with the structure of the actual execution plan that was used. For example, from an early version of 9i:
November 3, 2006
Did you know that it is possible for the cost-based optimizer to come up with a different execution plan simply because you changed the order of the tables in the from clause.
Although this can happen, in real-life, with real data and real indexes, it probably doesn’t happen often. Moreover, when it does happen you might not notice it because it’s likely to happen only in cases where the change doesn’t affect the performance of the query.