FreeLists / oracle-l / Re: why the cardinality of explain plan defers with the caridnality in row source operation

[Bob Carlin <smeghead.rimmer@xxxxxxxxx>]

Ajeet,
I would suggest looking at some tips at

http://structureddata.org/2007/11/21/troubleshooting-bad-execution-plans/

ajeet ojha wrote:
> Hello,
>  
> In my database environment for many of the queires the explain plan 
> cardinality defers (huge difference) with the cardinality shown by row 
> source operations of tkprof.
> I have been trying to find all possible reasons -
>  
> so far I came to know that 2 main reasons -
>  
> 1. limitation of cbo as mentioned in the link below
>  
> http://asktom.oracle.com/pls/asktom/f?p=100:11:0::::P11_QUESTION_ID:40230704959128
>  
> it also tells that solution is such cases is use dynamic_samping.
>  
> 2. the other reason is poor stats on the undelying tables/indexes.
>  
> please add if there are any more.
>  
> also how to collect a good stats  - is there any oracle recommendation 
> - best practices ?
>  
> are the above 2 are only reasons or there are many more - i went to 
> differnt forums and it seems that there are many more.
> please share if you have come across such sitution - possible 
> solutions etc.
>  
> Thanks
> Ajeet
--
http://www.freelists.org/webpage/oracle-l