Loading HALDBs that have secondary indexes

By default, when you perform an initial load of a HALDB database that has a secondary index, the secondary index is created as a part of the load process.

When source segments are loaded in HALDB databases, their secondary index entries are inserted in the secondary index partitions. The buffer pools that are used by the load program must include buffers for the secondary indexes. The secondary index partitions must be initialized before loading the database.

Initial loads do not create entries in the indirect list data set (ILDS), because all of the secondary index entries have accurate RBA pointers in their extended pointer sets (EPSs).

You can prevent secondary indexes from being created during an initial load of a HALDB database by specifying BLDSNDX=NO on the OPTIONS control statement in either the DFSVSAMP data set or the DFSVSMxx member of the IMS.PROCLIB data set. For example:

DFSVSAMP DD *
OPTIONS,BLDSNDX=NO
VSRBF=4096,500
IOBF=(8192,200)
/*

If you specify BLDSNDX=NO, you must build your secondary index by other means. You can use a tool, such as the IBM® IMS Index Builder for z/OS®.

When you use BLDSNDX=NO, the secondary index partitions are not authorized and their data sets are not allocated.

The use of BLDSNDX=NO and an index builder tool can shorten the load and index creation process, especially if you have many secondary index entries. When HALDB secondary indexes are created during the load process, the inserts in the secondary indexes are random, which can lengthen the load process considerably. Index builder tools like the IMS Index Builder read the loaded database, create index entries without writing them to the secondary indexes, sort the entries in secondary index key sequence, and then write them to the secondary indexes in a sequential process.