KMSYS Worldwide
All documentation
I-QU PLUS-1/User Guide

Database Reorganization User Guide

How to plan and execute DMS 2200 database reorganizations with I-QU PLUS-1, with worked examples and pointer-array handling.

Last updated August 10, 2011

Overview

I-QU PLUS-1 is a complete toolkit for reorganizing UDS DMS 2200 databases on the Dorado platform. It minimizes downtime by combining a flexible Q-Language unload/reload model with utilities that handle pointer correction, schema modification, and CALC simulation.

This guide walks through the reorganization process end to end: collecting schema information, building unload and reload runs, modifying the schema absolute, and running the reorganization itself. Three worked examples cover progressively richer scenarios, including pointer-array sets and global database pointer changes.

Why a dedicated reorganization tool

Conventional Unisys-supplied tools (DRU, FLDATA, etc.) work well for narrow cases but cannot handle changes in record length, restructured set linkage, or moving fields. I-QU PLUS-1 was built specifically to overcome those limits.

  • Handles changes in record length, field order, and set membership
  • Supports area-by-area unload and reload to keep the database online longer
  • Provides PBLD and PFIX to rebuild and repair pointers without reloading data
  • Includes SCHUTL for non-destructive schema absolute modification
  • Built-in CALC simulator removes the need for a live DMR during transformation

The reorganization process

Step 1 — Collect the information

Use QRYSCH to interrogate the schema absolute and dump all information needed to build the run streams: areas, page sizes, CALC chains, set ownership, and pointer layouts. The output (the SGSFILE) becomes the input to your unload, reload, PBLD, and PFIX runs.

Step 2 — Build the reorganization runs

  • Area unload — read each area sequentially, transform records, and write to a sort/output file
  • Area reload — read the sorted file back, allocate pages, and write records into the new schema
  • PBLD run — rebuild all owner-side pointers from scratch
  • PFIX run — repair specific broken pointers and finish set linkage
  • SCHUTL — modify the schema absolute to reflect the new structure

Step 3 — Run the reorganization

Take a current backup of every area involved before starting. Run unload, schema modification, reload, PBLD, and PFIX in sequence. Each step is restartable from its own checkpoint, so a failure mid-reorganization does not require starting over.

Note: A current backup of all areas involved must be taken before any reorganization.

Pointer arrays

Pointer Array (PA) and Indexed Pointer Array (IPA) sets need special handling because the owner record contains a packed array of database pointers rather than a chain. I-QU PLUS-1 provides a database key cross-reference and PBLD/PFIX modes that operate directly on PA/IPA structures so you can reorganize either the member or owner area without manually unwinding the array.

Global database pointer changes

When a reorganization changes the page size, area size, or area number of a database, every database pointer in the database has to be rewritten. Use the DBPCHG mode of PFIX to rewrite all pointers in place against the new schema absolute — no full unload/reload of unrelated areas is required.

Database record structure refresher

  • Record header word — record type, length, and prefix bits
  • Set ownership pointers — first/last member for each set the record owns
  • Automatic membership pointers — owner/prior/next for each automatic set
  • Manual control word and manual set pointers — connection state and links for manual sets
  • CALC and indexed sequential links — chain forward/back pointers for CALC and ISAM sets