Specifies the database where you are performing partition management.
table table-name
Specifies the source table for the PARTITIONMANAGE SPLIT operation..
partition table-partition-name
Specifies the partition instance being split.
composite initial
Specifies a special case for PARTITIONMANAGE SPLIT. "initial" is a reserved term indicating the name of a table's composite partition during data migration.
useindex index-name
Specifies the name of the index to use when scanning the table during the split. The index must be a local index. A non-unique index is allowed, but may adversely impact the performance of the PARTITIONMANAGE SPLIT operation.
res numrecs
The number of records to split in one transaction. If not specified, numrecs defaults to 100.
PARTITIONMANAGE SPLIT is used to redistribute the records in one table partition into different partitions. Use PARTITIONMANAGE SPLIT when a partition is too big, a partition needs to be partitioned into finer granularities, or for data migration when initially partitioning a table.
When a table partition is split, the records, LOB fields, and local index partitions of the table partition instance are moved to the appropriate target partitions, and global indexes are updated.
PARTITIONMANAGE SPLIT relies on the target partitions being previously defined. PARTITIONMANAGE SPLIT will allocate target partitions that are not allocated at the beginning of the split.
PARTITIONMANAGE SPLIT supports composite partitions, range partitions, or sub-partitions that have range at the end. Executing PARTITIONMANAGE SPLIT on list partitions is not supported, unless the source partition is a composite partition..
If execution of PARTITIONMANAGE SPLIT does not successfully complete, re-executing the command completes the split, from the last committed transaction.
PARTITIONMANAGE SPLIT is a resource-intensive operation.
If db-name is a UTF-8 database, you must specify -cpinternal UTF-8 on the command line. If not specified, an error message is generated and PARTITIONMANAGE SPLIT exits.