SMFD Control Statement Reference › SPLIT › Split File Time Processing

Split File Time Processing

In addition to creating raw split files with start and end times that match the history files, you can also create split files that cover specific intervals of time, where the start and end times are user-specified. The TIME operand of the SPLIT statement controls this, along with several other operands (all beginning with TIME).

Here are the similarities between TIME split file processing and non-TIME processing:

• Record selection/exclusion is identical.
• QSAM is used to write the records to the split files.

Here are the differences between TIME split file processing and non-TIME processing:

• Output split files must be part of a GDG for TIME processing.
• Split files are allocated and deallocated dynamically for TIME processing and must be indicated in the JCL for non-TIME processing.

To set up TIME processing for split files, you must do the following in advance:

• Set up GDGs that the split files will be members of, using Access Method Services.
• Determine the times of day (up to 4) when the split files will be separated.
• Define appropriate SMS storage class definitions if SMS is to be used for allocating the split files.

For more information, see the Systems Programmer Guide.

TIME processing of split files requires an index file. The TIME process uses the index file to complete files that were started at the end of a dump process. For instance, suppose that the value of the TIME operand is TIME(0000,0800,1600), and an SMF dump process begins at 4:00 a.m. for 5 hours' worth of data. The dump process will find the split file that was begun at 1600 the previous day and add the records in this dump from 2300 the previous day until the date changes. At the point of the date change (time becomes 0000), the current split file will be closed and deallocated. The index will be updated and a new record added for the just-closed split file indicating that the file is complete. The dump process will then allocate a new generation of the GDG to be the new split file and open it for processing of additional records. When the process is complete, an entry will be added to the index indicating the current split file is incomplete. The next dump process will then be able to find the split file and add more records to it until records for 0800 or later are found. Also, any postprocessor of the split files will be able to determine if the file is ready for processing.

Note: While some management of the split files will be done via the index and the TIME operand, there are no additional controls placed on the split files or on the index. The index is a character-based sequential file, so it can be manipulated with almost any text editor on the mainframe. Be sure that users and applications have appropriate access levels set for the index file and the split file GDSs. Do not attempt to access an incomplete time-managed split file (that is, generation 0).

If other applications need to process the information that is found in the split file index, the INDEXDUP operand should be coded on the SPLIT statement in order to generate a copy of the index for use by other applications.

More information:

Using Split Files Generated with the TIME Operand