Changing compression techniques for full-function databases by using compressed data
When using the IBM HD Reload utility and changing compression techniques before expanding the data, it is important when (before or after the unload) and how you change the DPICDS with the new technique information.
To change compression techniques while your data remains in a compressed format, complete the following steps:
- Make an image copy of your database and use IEBCOPY to back up your DPICDS and DPITBL data sets before starting any changes.
Generate a Reorg DBD, needed in the unload and reload steps.Use the following control statement to generate the Reorg DBD:
BUILD DBD(name),SEG(&ALL(LEAVECOMPRESSED))For more information about the BUILD command, see Reorg DBD utility—overview.
Unload the database with the Reorg DBD created in Step 2.You can use the IBM HD unload utility. Specify the Reorg DBD in your IMS DD statement as follows:
//IMS DD DSN=DPIMS.REORG.DBDLIBRun the Trial utility by using either the image copy created in Step 1 or by using the database data set as input. (required for changing to Custom Huffman or to a new Shared Table or Custom Hardware Dictionary; otherwise optional)You can run the Trial utility to determine the technique that produces the highest compression.
To change to Custom Huffman, or to change to Shared Table or Custom Hardware Dictionary (unless you use an existing table or dictionary), you must run the Trial utility to build a new table or dictionary and to register it at the STAGE level.
Change the DPICDS.For details, see Changing-the-DPICDS.
For each segment you change to or from the Custom Huffman, Shared Table, or Custom Hardware Dictionary techniques, make sure the DPICDS has the tables or dictionaries required for the original technique and for the new technique. Data Packer determines the original technique from an integrity byte in the segment data and looks for the technique information during the reload - first at the RUN level, then at the ARCHIVE level.
Reload the database with the Reorg DBD created in Step 2.You can use the IBM HD Reload utility.
Special statements for the DBD libraries must be included. You must include the following DBDLIBs in your reload JCL:
//IMS DD DSN=DPIMS.V2.DBDLIB(DBD with DPIEXIT on the segments)
//$$DPIDBD DD DSN=DPIMS.REORG.DBDLIB(the Reorg DBD created in Step 2. This special DD statement alerts Data Packer to expand data with the old technique before it compresses the data with the new technique.)