Tuesday, February 21, 2012

Attributes Change Run - applying changes in aggregates

Basics of Attribute Change Run

Attributes Change Run applies changes into aggregates if there was a change in navigation attributes used in aggregates or in hierarchies of a characteristic.

This process in necessary, because in aggregates data from InfoCube are persistently stored in aggregated form (In InfoCube, no data related to navigation attributes or hierarchies are stored).

To start Attributes Change Run, go to:

RSA1, use menu Tools>Hierarchy/Attribute Changes.

You will see there a list of available sets of InfoObjects and appropriate change status.

If you mark a line, and click InfoObject List button, you will see a list of objects that are planned for that Change Run.

The second important button is Execute button. Use it when you already market the line, for which you would like to start Attribute Change Run.

Change Run process steps

  1. Change Run checks which characteristics and hierarchies had changes since the last change run. Run specifies which aggregates must be adjusted.
  2. Change Run apply changes (dialog process for each required InfoCubes' aggregates).
  3. Activation of changes in master data and hierarchies.

There is a SAP lock, that (during point 1) is preventing other changes runs to start a job for the same time. Also rollups and aggregate builds are not executed. The name of the lock: CHNGRUN_ST

The second SAP lock (CHANGERUN) is locking objects (InfoObjects, InfoCubes, hierarchies) that are involved in the adjustment (point 2).

When the work is finished, the CHANGERUN lock is removed.

There could be only one change run in the initial phase (1). If other change run starts, it will fail immediately. It is not possible to start multiple change runs. Read about parallel change run in SAP Note 534630.

Additional Notes:

Change Run process can be monitored, restarted using RSDDS_CHANGERUN_MONITOR ABAP report. You can start it using se38 transaction.

It is better to run CHR for all newly loaded master data in the process chain. Do not start it for each master data load.

Usefully links:

No comments: