Tuesday, March 06, 2007

Hotso 2007 Day 2

Day 2 started with a warm and fuzzy feeling.
started off with a presentation on profiling and how 10046 can tell you stuff awr ,addm etc wont tell you.
there was a mention about OraSrp.
I had heard about the tool but not used it. I will definitely try to use this and see how good this is.
one of the things that i didnt know was the fact that <OCI8 Bind peeking is not done. See you always learn something new everyday.
Jonathan Lewis's paper was about Interpreting Statspack reports and the knowledge imparted was very good.
I learned that index block splits in bitmap index could cause a log file sync since there is commits involved specially if deletes happened ( this is my interpretation of what he said , Please correct me if i am wrong).
Also hard pars=optimize so hard parses are not always a bad thing.
Joze Senegacnik presented his paper on ATO and sqltuning features in 10.
This was nice and interesting topic since it had nice content.
stuff i need to look into are the ignore_optim_embedded hint and the opt_estimat hint.
sql profiles do look good but they might pose a problem too.
Tanel Poder had back to back presentations but i only attended the scalability in 10&11g.
it was nice to hear about the new mutex architecture. the buffer cache/shared pool sharing.

James Morle had the Brewing Benchmarks Presentation.
the tool he discussed can be downloaded under GNU license here.
The tool does look promising and i will download and do a small load test to see how it works.
He did talk about a bug which i might want to look at about Pl/Sql RPC call not being logged to 10046 traces till 10g.This info is available in 10051 trace
there is a once off patch for 9i and .

There was a speaker panel with interesting questions.
I was planning on attending the oracle-l dinner but at the last moment had a bad headache and came back to my room.

Google Search