Considerations for HALDB partitioned secondary indexes

A secondary index of a HALDB database must be a HALDB partitioned secondary index (PSINDEX).

A PSINDEX can have one or more partitions. The key ranges for the partitions of a secondary index are not likely to match the key ranges of its target database. Partition selection for the secondary index is based on its key; partition selection for the target database is based on its key. Usually, these keys are unrelated. In some cases, you can use a partition selection exit routine to partition the secondary index along the same boundaries of the main database. An appropriate key is required, because the selection of a partition always depends on the root key. The partition selection exit routine uses the parts of the root key that you specify to select the correct partition.

To initialize partitions in a PSINDEX, use the HALDB Partition Data Set Initialization utility (DFSUPNT0). DFSUPNT0 automatically generates recovery points for the PSINDEX. Recovery points are not created if you delete and redefine your PSINDEX partitions and then turn off their PINIT flags.

Additionally, there are other restrictions and requirements for HALDB partitioned secondary indexes:

• Symbolic pointing is not supported.
• Shared secondary indexes are not supported.
• Secondary indexes must have unique keys.
• /SX and /CK fields can be used to provide uniqueness.
• The reorganization of a target database does not require the rebuilding of its secondary indexes. HALDB databases use an indirect list data set (ILDS) to maintain pointers between the PSINDEX and the target segments.