History data in optimization processing

For example, when a VSAM ACB specifies both sequential and direct processing, there is no way to know which type of access predominates, or whether one is used at all. For this reason, Data Optimizer collects historical information during close processing, when appropriate, for future reference. At the next open, Data Optimizer uses the information to make informed tuning decisions. In this way, Data Optimizer learns the application access patterns to improve performance.

Data Optimizer has two methods to relate historical data to the file being opened: DSNPGM and FULL. Data Optimizer provides the HISTKEY command to determine which method to use when creating the history. For more information on the HISTKEY command, see Data-Optimizer-commands.

• The DSNPGM method uses the program name and data set name combination to access the data.
• The FULL method uses job name, step name, procstep name, program name, and ddname to access the historical data. This is the default method.

If history has been recorded using the DSNPGM method, that data is used to prime the decision-making process for the FULL method. Priming of historical data ensures that the buffering decision is consistent.

To view the history information that Data Optimizer has collected, generate the Data Optimizer History Report. For more information about this report, see History-report.

To create a sequential data set backup of the BMC AMI Batch Optimizer  History data set, customize and execute member BSSCJEXT of the SAMPLIB data set. The BMC AMI Batch Optimizer  Subsystem must be executing when this utility is submitted. The sequential data set that is created by this utility can be used as input when IDCAMS is executed to recover the History data set.

BMC recommends that you periodically reorganize the History data set. Member BSSREORG of the SAMPLIB data set uses IDCAMS to complete an EXPORT, DELETE, DEFINE, and IMPORT of the History data set.

History cleanup utility

The history cleanup utility removes records that belong to the Data Optimizer component. This utility removes data that becomes obsolete over time. 

The utility's JCL, which is in prefix.XXSAMP in member DAPUCLN1, does not directly update the history registration data set (REGISET).  The utility creates a sequential data set that you can use as input when executing IDCAMS to recover the History data set from prefix.XXSAMP in DAPUCLN2.

Before executing member DAPUCLN1, ensure that the BMC AMI Batch Optimizer  subsystem is shut down to avoid losing data that is being updated in the history REGISET.

When you are sure that correct records are deleted, make the necessary changes to member DAPUCLN2 before submitting the JCL for execution.

The Data Optimizer uses control cards that direct the history cleanup utility to delete the history data. The utility does not write the deleted data back to the History data set.

The RXSYSIN DD statement points to the control cards for deleting history data.

The following table displays the parameters in the history cleanup utility's summary report: 

The following example shows deleted records and the history cleanup summary for the Data Optimizer:

To run the history cleanup utility

1. Shut down the BMC AMI Batch Optimizer  subsystem by determining the subsystem identifier (the default is MBOS) and issuing the SHUTDOWN command.
   
2. Customize the member DAPUCLN1 in the XXSAMP data set. DAPUCLN1 provides customization instructions.
3. Execute DAPUCLN1.

4. Examine the REPORT DD statement in DAPUCLN1 to verify that utility deleted the correct data sets.

   Warning

   Note

   The history data remains in the data set until you execute DAPUCLN2 in the next step.

5. Execute DAPUCLN.

   Warning

   Note

   This job completed the following actions:

   • Deletes your existing REGISET data set
   • Defines your new history data set
   • Takes records from the sequential data set created in DAPUCLN1
   • Copies the records (REPRO) from the sequential data set and writes them in the history data set
6. Start the BMC AMI Batch Optimizer  subsystem.