gtpm1m3s | TPF V4R1 Migration Guide: 3.1 to 4.1 |
This section discusses database reorganization (DBR) and topics such as:
See TPF Database Reference for more information about DBR.
In the TPF 3.1 system, database reorganization (DBR) used the CREDC macro to write output records to tape. Each CREDC macro was associated with one block.
In the TPF 4.1 system, DBR uses ECBs, but the ECBs are associated with frames (in large numbers). Before you run DBR, you need to make sure that you have allocated enough frames or you will run out of main storage.
Every environment is different, and numbers may vary significantly, but a guideline might be to allocate 20 frames for an ECB. The number of ECBs defaults to 4 (on the ZDBRO command), so if you allocate (4 ECBs) × (20 frames) = 80 frames for DBR, that should be sufficient. Because you need frames for other activities as well, you should not allocate more than one-third of the frames in your system to DBR. You should allocate less than this if you run DBR on more than one subsystem or subsystem user simultaneously, or if there is more activity in your system. Also, you should use the ZDBRO OECB xx command (where xx is the number of ECBs to be used for output) cautiously because increasing the number of ECBs increases the number of frames needed for DBR.
When you use the ZDBSO and ZDBSI commands to capture the entire database, you now have individual control on processor and I-stream unique records.
There is a new option on the ZDBSO INIT command so that you can selectively specify a group of subsystem users (SSUs). This option initializes the database reorganization (DBR) control records on only the SSUs you specify. You can initialize the control record with record types to be captured for a given SSU, processor, or I-stream with the ZDBRO INIT command.
If you want to run DBR on all shared records, enter the ZDBRO INIT command from a single SSU, processor, or I-stream. If you have mostly shared records with a few unique ones, run DBR after entering the ZDBRO INIT command from a single SSU, processor or I-stream, then run DBR by entering the ZDBRO INIT command to capture all the unique records owned by the SSU, processor, or specified I-stream. If you have a lot of unique records, run DBR using the ZDBSO and ZDBSI commands.
In the TPF 4.1 system, the first pool record type is assigned to the next value after the last fixed record type. In the TPF 3.1 system, pools were assigned record types 512 through 520 on all systems. In the TPF 4.1 system, DBR was changed to handle the new pool record type values with little effect to you.
In the TPF 4.1 system, the number of record types that can be defined in the system was increased to 12 KB.
In the TPF 3.1 system, all records captured by the DBR output phase were read in during the DBR input phase.
In the TPF 4.1 system, only those records whose record ID attribute table (RIAT) attributes are setup as RESTORE=YES on the RIAT macro are restored to the new system during the DBR input phase.