6.1.5 Loading a large amount of data (data loading with the synchronization point specification)

If you plan to load a large amount of data to a table, evaluate the use of data loading with the synchronization point specification.

Normally, with data load processing, a transaction cannot be settled until all data store processing is completed. Therefore, a synchronization point dump cannot be validated while the database load utility is executing. If HiRDB terminates abnormally while loading a large amount of data, it takes a long time to restart HiRDB. To avoid this, you can set a synchronization point at any interval based on the number of data items during data load processing, thereby enabling transaction settlement. This is called a data load with the synchronization point specification.

To perform data loads with the synchronization point specification, specify a line number of a synchronization point (the number of data items after which a synchronization point is to be set) using the database load utility's option statement.

Notes
  1. When this facility is used, the overall throughput is reduced by the synchronization point processing, compared to when this facility is not used.
  2. If the utility terminates abnormally, the recovery method depends on the termination timing. For details about how to handle abnormal termination, see 6.7 Handling utility abnormal termination errors during data loading with the synchronization point specification. Note that the recovery method is complicated if the utility terminates abnormally during a data load in the batch index creation mode.
  3. Because data storage begins on a new page for each synchronization point, this facility is applicable only when a large number of pages is stored.