Overview of Parallel Scan Facility

The FABHFSU Parallel Scan Facility (PSF) makes possible a significant reduction in the amount of elapsed time required to scan multivolume databases. It does so by scanning separate, predefined portions of the database simultaneously.

A PSF run results in multiple output data sets that must be combined to represent the output of a complete scan. PSF supports only HDAM and HIDAM databases; PHDAM and PHIDAM databases are not supported.

The FABHFSU utility in PSF mode consists of the following four programs:

FABHPSFM

Provides assistance to the user in determining how to divide the database into portions to be scanned by PSF phases.

FABHPSFC

Processes the control statements that define the scan and builds a scan control data set that is used by the remaining functions to obtain and record information about the scan.

FABHFSU

In the PSF mode, FABHFSU performs individual scan phases, which can run separately or concurrently to unload a multivolume database. In the PSF mode, the information is obtained from the scan control data set created by FABHPSFC.

FABHPSFS

Creates summarized statistics and data set concatenation sequences. For an unload format, prepares the header and trailer records required. This is the last program to be run.

Because all facilities provided for a typical FABHFSU run are available for individual PSF scan phases, PSF can benefit as follows:

• Earlier detection of severe errors that must not be bypassed. This is especially advantageous when the error is toward the end of the database or if there are multiple errors.
• Reduced recovery time from database errors. This is possible because only the affected scan phases need to be rerun. Therefore, other scan phases (even those in which the beginning node is located beyond the point of error in the database) can continue to run while the errors are being analyzed or corrected.
• Reduced recovery time from a permanent write error at unload time or a permanent read error at reload time. This is possible because only affected scan phases need to be rerun. If dual outputs are being created, this would not be relevant unless both formats encountered permanent errors or damages.