Features and Benefits
IBM DB2 High Performance Unload for z/OS (also referred to as DB2 HPU) is a high-speed DB2 utility for unloading DB2 tables from either a table space or from an image copy. Tables are unloaded to one or more files based on a format that you specify. DB2 HPU helps you manage and control the unload activity. It works outside DB2, directly accessing the VSAM or sequential files that contain the table space or image copy data set. DB2 HPU also provides superior performance, especially in terms of CPU and elapsed time.
By using DB2 HPU, you can alleviate some of the problems that are associated with unloading large amounts of data. Large sequential reads of DB2 tables take a long time. Because of this, the large scans that are required for unload are hard to schedule in the ever-shrinking batch window at DB2 installations. Performance becomes critical when several unloads read the same table space concurrently. The associated DB2 buffer pool management can cause multiple programs to compete for the same data, resulting in:
Conflicts in the use of the DB2 buffer pool
Writing over buffers that might be serving several unloads
Multiple reads of the same DB2 pages
Potential channel conflicts DB2 HPU relieves these problems. It provides an extremely fast way to sequentially read and share a DB2 table space among multiple unloads. DB2 HPU scans a table space and creates the number of output files that you specify, in the format that you specify. The output format can be any one of the following types:
DSNTIAUL compatible
VARIABLE, which lets you quickly create variable-length records
DELIMITED, which lets you quickly create a delimited file that you can export to another platform
USER, which lets you specify virtually any type of conversion so that your output appears as you want it
EXTERNAL, which lets you quickly create fixed length human-readable formatted files
Whenever possible, DB2 HPU processes requests to unload data from the same table space in parallel. You can create different output files during the same unload process at almost no additional cost. For example, you can unload a list of customers who have payments due this week and another list of customers whose birthdays are on the first day of the week. You can create these lists in a single execution of DB2 HPU at a fraction of the cost that is required by traditional dual unload executions. You can also run DB2 HPU against image copies of the table space, thereby avoiding interference with DB2 production databases.
To maximize performance, DB2 HPU uses buffering, synchronization techniques, and parallel processing.
Buffering - When reading data rows, DB2 HPU directly accesses the VSAM clusters that contain the table space. This direct use of VSAM takes maximum advantage of the buffering capability that is provided by VSAM, which means that an entire cylinder can be read with a single I/O.
Synchronization - DB2 HPU permits the parallel execution of several unload jobs that access the same table space; it does this by synchronizing the unloads.
Parallel Processing - Unload requests that work on the same table space are processed in parallel whenever possible. Unload requests that work on different table spaces can be processed in parallel if they are specified in the global parallelism setting.