Re: 11.2 change in direct insert behavior
Date: Sun, 26 Feb 2012 19:49:46 +0000 (UTC)
Message-ID: <pan.2012.02.26.19.49.46_at_gmail.com>
On Sun, 26 Feb 2012 10:46:35 -0800, Charles Hooper wrote:
> From the above, it should be clear that the APPEND hint had no effect on
> the execution plan, while the APPEND_VALUES hint caused the execution
> plan to change. We end up with a mode 3 TM lock on the table.
Which is a standard table lock issued with every transaction, "shared row exclusive". It is much, much better than mode 6. Actually, it's twice as good because 3*2=6 :)
>
> The changes are documented in the following Metalink (MOS) articles: Doc
> ID 842374.1 Doc ID 1391897.1
Very, very interesting. Thanks, Charles. I'm your fan ever since reading the chapters about performance and stats management that you wrote with Wolfgang Breitling in that furniture store book in 2010. Joze Senegacnik's chapter about extending CBO and creating proprietary cost function is also very interesting but a bit of a bleeding edge for me. I'm still looking for a good place to apply it. I used to work with Joze for the same company, a SGI reseller. I also used to work with Arup Nanda, for Oxford Health Plans. This is a small world.
-- http://mgogala.byethost5.comReceived on Sun Feb 26 2012 - 13:49:46 CST