Comments on: Correlation, nocorrelation and extended stats http://orainternals.wordpress.com/2008/12/19/correlation-nocorrelation-and-extended-stats/ Discussions about Oracle performance tuning, RAC, Oracle internal & E-business suite. Sun, 19 May 2013 12:13:20 +0000 hourly 1 http://wordpress.com/ By: Alberto Dell’Era’s Oracle blog » CBO: NewDensity for Frequency Histograms,11g-10.2.0.4 (densities part IV) http://orainternals.wordpress.com/2008/12/19/correlation-nocorrelation-and-extended-stats/#comment-421 Fri, 23 Oct 2009 16:08:06 +0000 http://orainternals.wordpress.com/?p=182#comment-421 [...] the "Histogram change" post by Jonathan Lewis, which distills the findings of Randolf Geist and Riyaj Shamsudeen) – a surprising rule that might easily cause trouble (in fact as Jonathan reports in the comments – [...]

]]> By: Randolf Geist http://orainternals.wordpress.com/2008/12/19/correlation-nocorrelation-and-extended-stats/#comment-195 Sun, 25 Jan 2009 00:13:07 +0000 http://orainternals.wordpress.com/?p=182#comment-195 Riyaj,

I’ve only recently come across your interesting blog post regarding correlated column values.

I think that some of your findings were significantly influenced by the frequency histogram that you’ve generated. I’m not sure why you did so, since there is no data skew.

In particular the selectivity results of the “index + histogram” case (cardinality => 200) and the “extended stats + histogram, out-of-range predicate” case (cardinality => 499) seem to be mainly driven by the fact that there are frequency histograms on the (virtual) columns.

I’ve put together some amendments to your test case and published them on my blog which illustrate the results with and without histograms, and in case of histograms, the influence of the “bug fix” Bug 5483301 – Cardinality of 1 when predicate value non-existent in frequency histogram, which is described in Metalink Doc ID 5483301.8:

http://oracle-randolf.blogspot.com/2009/01/correlation-nocorrelation-and-extended.html

Best Regards,
Randolf Geist

]]> By: orainternals http://orainternals.wordpress.com/2008/12/19/correlation-nocorrelation-and-extended-stats/#comment-180 Fri, 16 Jan 2009 14:13:02 +0000 http://orainternals.wordpress.com/?p=182#comment-180 Hi Chris
Thanks for reading my blog and Thanks for sharing your observations about that parameter. I researched about that parameter too, but I couldn’t gain anymore insight in to it. It is excellent that you are able to decipher meaning of this parameter value.

Thanks!

Ciao
Riyaj

]]> By: Christian Antognini http://orainternals.wordpress.com/2008/12/19/correlation-nocorrelation-and-extended-stats/#comment-177 Thu, 15 Jan 2009 10:48:34 +0000 http://orainternals.wordpress.com/?p=182#comment-177 Hi Riyaj

I do not know how _optimizer_extended_stats_usage_control works… anyway, two observations I did:
- The value changed from 240 (11.1.0.6) to 224 (11.1.0.7). IN other words, they disabled the fifth bit.
- When the second bit is set, extended stats seam to be disabled.

Cheers,
Chris

]]>