Performance tuning by type of input data

If you are loading data in certain environments, the type of input that you provide can affect the performance of your load job.

Loading data from UNLOAD PLUS (FORMAT BMCUNLOAD option)

When used in conjunction with UNLOAD PLUS, LOADPLUS provides a high-speed option to move data from one table to another table that has a similar table structure. This feature is useful, for example, for migrating data from test to development databases. To unload the data, you specify the FORMAT BMCLOAD option in UNLOAD PLUS. UNLOAD PLUS unloads the data in an internal format that only LOADPLUS can use. To load the data, you specify the FORMAT BMCUNLOAD option on your LOAD command.

Because the data is never converted to an external format and because LOADPLUS does not perform data verification, using this option can help reduce CPU cycles and elapsed time for your load job as well as for the entire process of migrating data.

Loading XML data

When loading XML data, adjust the XMLAVGSIZE global option or AVGSIZE field option to improve the performance of your load job.

Loading LOB data

When loading LOB data, adjust the LOBAVGPCT global option or AVGSIZE field option to improve the performance of your load job.

Loading data from a single input file

When loading data from a single input file and the data is already sorted, specify ORDER PRESORTED to obtain optimal performance.

Loading data from multiple input files

When you are loading data from multiple input files, use multiple SYSREC data sets. LOADPLUS can run three read tasks per active CPU in an LPAR. The optimal configuration is to distribute the amount of data evenly across all of the SYSREC DDs. In addition, you can concatenate multiple data sets to a single SYSREC. Use the following guidelines for setting up specific types of load jobs:

If one file has half of all of the data and the rest of the data is in more than one additional file, allocate the one large file to one SYSREC, and the rest of the files concatenated to a second SYSREC. The input process is never shorter than the longest task, so adding additional SYSREC files does not decrease the elapsed time and uses additional resources that could be allocated more productively.
If you have 16 data sets and want to run four input tasks, specify SYSREC01 through SYSREC04 and have four data sets allocated to each SYSREC nn.

For requirements when using multiple SYSREC data sets, see SYSREC-data-sets-in-LOADPLUS.