Use this option when you want to turn on or turn off the "high used mark" option.
When this option is not turned on, each execution of a DBUTLTY LOAD loads the data (if any) then formats the rest of the data set by writing binary zero blocks. It is possible, especially at sites using DASD mirroring, for these formatting writes to overwhelm the system, thereby causing possible performance degradation. Also, without this option turned on, physical retrieval against an area, such as the GETPS command and, for example, the DBUTLTY function BACKUP SEQ=PHYSICAL, requires the entire data set be read even though much of it may never have been used.
After reading the following information, you can decide if this feature is desired at your site. If so, you can turn it on for all data sets in the Directory (CXX). Once turned on, some existing programs and DBUTLTY jobs might execute in less time, because thereafter less than the whole data set area is loaded and (or) read. It is therefore possible that significant performance gains can be realized for sites that frequently load their data or use the GETPS command in programs.
After this option is turned on, the first execution of the DBUTLTY function LOAD after a DBUTLTY function INIT of an area continues to function as it did before this option was turned on. All subsequent LOADS, however, including REORG, and physical retrievals (GSETP command and BACKUP SEQ=PHYSICAL) rely on a special form of the "next available overflow" called the high used mark feature.
The next available overflow is a relative DASD location, part of the data set that has never been actively used by CA Datacom. The high used mark represents the point up to which LOAD must format the data set and up to which physical retrieval must read. It is the next available overflow number rounded up to 64 tracks.
An INIT of a data area sets the high used mark to zero. A value of zero means LOAD must format the entire data set. At the start of LOAD, the current high used mark is saved and the DASD version is set to zero. If the data LOAD takes less space than the saved copy of the high used mark, the data set is formatted up to that mark. After the data portion of the LOAD is finished but before the index portion of the LOAD is done, the DASD high used mark is updated to indicate the point to which "true data" has been loaded, rounded up to a 64-track boundary. If the LOAD fails during the data portion, the next LOAD must format the entire data set. If the LOAD fails during the index portion, the high used mark is used for formatting.
The runtime code in the MUF that maintains the next available overflow notices each time the next available overflow number crosses a 64-track boundary. If the memory copy of the high used mark is non-zero, MUF reads the data set control block. It then updates the high used mark and immediately writes the data control block.
A GETPS command treats the high used mark as the end of the data set unless the area is not opened for update. If it is not open for update, GETPS reads to the true end of the data set. This is done for the sake of read-only MUFs. LOAD always treats the high used mark as the end of the data set.
If you turn off this option, it reverses the marking of the CXX caused by turning the option on. When an area is opened for update, its compatibility level is checked. If it is level one and the CXX has been marked to stop using the high used mark feature, the compatibility level is reset to 0, the high used mark is zeroed.
|
Copyright © 2014 CA.
All rights reserved.
|
|