DB2 Version 9.7 for Linux, UNIX, and Windows

FP1: Distribution statistics collected for XML columns

Starting with DB2® Version 9.7 Fix Pack 1, distribution statistics can be collected for XML columns to support faster queries over the data in XML columns. Distribution statistics are collected for indexes over XML data of type VARCHAR, DOUBLE, TIMESTAMP, and DATE.

For DB2 V9.7 Fix Pack 1 and later, distribution statistics on an XML column can be collected.

Distribution statistics are collected for indexes over XML data of type VARCHAR, DOUBLE, TIMESTAMP, and DATE. XML distribution statistics are not collected for indexes over XML data of type VARCHAR HASHED.
Distribution statistics are collected for each index over XML data specified on an XML column.
XML distribution statistics are collected when automatic table RUNSTATS operations are performed.

To collect distribution statistics on an XML column using the RUNSTATS utility, both distribution statistics and table statistics must be collected. Table statistics must be gathered in order for distribution statistics to be collected because XML distribution statistics are stored with table statistics. As the default, the RUNSTATS utility collects a maximum of 250 quantiles for distribution statistics for each index over XML data. The maximum number of quantiles for a column can be specified when executing the RUNSTATS utility.

The following list describes the situations in which XML distribution statistics are not created or collected:

XML distribution statistics are not created when loading data with the STATISTICS option.
XML distribution statistics are not collected for partitioned indexes over XML data defined on a data partitioned table.
XML distribution statistics are not collected when collecting only index statistics, or collecting index statistics during index creation.