now I understand that my wording was little bit confusing, thanks. By “suited” I mean they are using them and I consider all parties are happy with what they are using.

I’ve seen way too many instances of moving to the defaults helping things to assume it’s all being done for valid reasons.

Lucky you

Oracle has worked to address the issues with 10g stats defaults (mainly sample sizes and bind peeking), so the messaging (and my practice) is to re-evaluate and reconsider when going to 11g.

I agree that auto sample size in 11g rocks – one of the best improvements in 11g overall I think. However I don’t think bind peeking issues have been worked out on the same quality level. In my view they just help to workaround root cause (sudden histograms), not fix it. Sure they will help in some other cases too, but they are not interesting to me. My opinion is: it’s better not to have a histogram in place, than have some, and Oracle’s default behavior is exactly opposite.

If things are not working, then feedback into the product teams is necessary to help advance the product.

Unfortunately many people don’t know that Oracle is using bugfeedback-driven development, so many cases remain unexposed.

Second, Oracle has worked to address the issues with 10g stats defaults (mainly sample sizes and bind peeking), so the messaging (and my practice) is to re-evaluate and reconsider when going to 11g. Given your poll, there is no way to detect trend to see if that is happening.

Again, I think it is important to point out that Oracle does not expect the defaults to work for all cases, but there certainly should be more cases that the defaults work for in 11g, than 10g, and it should continue to be that way with future releases.

If things are not working, then feedback into the product teams is necessary to help advance the product.

thanks for your comment

This might be jumping to conclusions. The poll does not contain version numbers with each vote, and I suspect that the 11g defaults would work for far more installations than the 10g ones did

I don’t think this is jumping to conclusion – it’s what poll results report. The fact that defaults would work better in 11g doesn’t change a thing if they are not used. My opinion is if somebody doesn’t use defaults in 10g, chances are high they wouldn’t use them in 11g too. I’m talking from point of view of a big software vendor that uses DBMS/AS as a tool and have to make software work properly, so I might be wrong – my view may be biased towards not relying on Oracle’s out-of-the-box automatic features. Plus I’ve seen far less installations than you.