db-derby-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Mike Matrigali <mikem_...@sbcglobal.net>
Subject Re: Question on why indexStat deamon is being triggered
Date Fri, 30 Mar 2012 16:06:18 GMT
Kristian Waagan wrote:
> I'd say a threshold of 4.0 is pretty relaxed - do you expect the 
> distribution/cardinality of the values to be similar for a table with 
> 50K rows and 2500K rows?
In Derby statistics are only used for cardinality estimates in the
optimizer.  They are not used for distribution estimates.  Distribution
information is taken from the indexes themselves and thus are zero
admin, always maintained.  I just keep noting this as this is very
different from other databases.

I believe the cardinality stats are such that if on average on a
single column index and 100000 rows with an average of 10 duplicates
per value, and then you grow to 1 million rows the existing statistic
will assume 100 duplicates per value.

View raw message