Performance considerations for the INTO and UNIQUEINTO command options

UNIQUEINTO YES tells LOADPLUS that as soon as an INTO specification selects an input record, LOADPLUS should not test the record against other INTO specifications.

Normally, LOADPLUS checks each input record to determine if it matches the WHEN or PART criteria of each INTO specification. Checking starts with the first INTO statement and ends with the last INTO statement. This can be extremely time-consuming when you specify many INTO statements.  

INTO and PART

Use the following guidelines to determine how many INTO statements to use with your job and when to use the PART option for optimal performance of your job:

• Avoid specifying a separate INTO statement with PART option for each partition.
• When either of the following conditions exists, specify only one INTO statement without the PART option. The resulting partitioning key determines the partition into which the record is loaded, so the PART option is not necessary.
  • If you specify LOAD REPLACE
  • If you specify LOAD RESUME YES and are loading the majority of partitions in a partitioned table space
• If you specify LOAD RESUME YES and are loading only a few partitions, specify all partitions on one INTO statement (for example, INTO PART 1,3:5). This specification eliminates unloading and loading all partitions of partitioned indexes and can save considerable time. 

UNIQUEINTO YES

Use UNIQUEINTO YES if both of the following conditions exist:

• You are loading multiple tables of a nonpartitioned table space or multiple partitions of a partitioned table space.
• Each input record matches at most one INTO specification.

As soon as an input record matches the WHEN selection criteria of an INTO statement, LOADPLUS does not check any more INTO specifications. For each input record, this reduces the number of INTO specifications that LOADPLUS checks to half of the INTO specifications made, on average, which can improve performance significantly. You can also improve performance by specifying the INTO statement that selects the most input records, followed by the INTO statement that selects the second most input records, and so on.