Data Partitions

Data partitions allows you to split large tables into logical partitions.

For maximum performance, run each logical partition in parallel on a database with multiple engines. You can also run the data partitions in smaller groups, if DA is configured to run fewer comparisons at the same time.

Note: When you run a comparison with a new compareset for the first time, DA runs the comparison using a single partition, and collects partition information for subsequent runs.

The SQL where clause on the compareset key columns defines the upper and lower partition boundaries. The boundary key values are sampled for a compareset when a comparison that uses it is run for the first time. These boundary samples are stored in the DASD.

On subsequent runs, any comparison that uses the same compareset will use those boundary samples to split a table into many logical partitions.

For example, if there are 16 boundary samples stored and the NUM_PARTITIONS comparison option is set to 2, only the "middle" boundary sample is required to split the table into two logical partitions.

Each time a comparison is run, DA collects new boundary samples for its compareset. The new boundary samples for the compareset in use are updated in the DASD.