When you use a batch job distributed execution system to divide data and execute batch processing, you can process batch jobs even faster by increasing the number of parallel executions.This is done by adding units for data storage. A unit that can be added to increase the number of parallel executions while HiRDB is active is called a HiRDB expansion unit. Once batch processing is finished, you can either simply terminate HiRDB expansion units or delete them from HiRDB.

A unit that is running before a HiRDB expansion unit is added is called a base unit.

The following figure shows base units and a HiRDB expansion unit.

Figure 29-2 Base units and a HiRDB expansion unit

Explanation

During normal operations, only the base units are active, but HiRDB expansion units can be added to HiRDB and started as needed. HiRDB expansion units are defined in the pd_system_expand_unit operand of the system common definition in advance, to set up the environment before actual operations begin.

When a HiRDB expansion unit is started, it is treated like HiRDB.

When a batch job distributed execution system executes a grid job, a grid job pre-processing program collects the placement information of the data stored in HiRDB. This information is used to control execution of jobs by the HiRDB expansion units that have been added.