Archiving Data from Non-Cumulative InfoCubes 

Use

In order to archive data from InfoCubes with non-cumulative values, you need to be aware of a number of peculiarities in the procedure.

For more information on the structure of InfoCubes with non-cumulative key figures, see Non-Cumulatives.

Features

There are two ways of archiving data from a non-cumulative InfoCube - time slot archiving and archiving with selection characteristics.

Time slot archiving (recommended)

As an archiving method, you must choose for a non-cumulative InfoCube, time slot archiving for the time reference characteristic. When you recreate an archiving object for a non-cumulative InfoCube, this characteristic is also suggested by the system.

Only archive non-cumulative changes up to a specific point in the past. This reduces the volume of data in the InfoCube. The marker (= end non-cumulative) remains unchanged.

In the variants maintenance screen, in the archive administration, under Non-Cumulatives, choose the Movements Only option.

You are only able to archive the marker if you also archive all of the non-cumulative changes. Subsequently, the contents of the non-cumulative InfoCube are archived completely. In the variants maintenance screen, in the archive administration, under Non-Cumulatives, choose the Movements and End Non-cumulative option.

You must archive the non-cumulative changes (for all times) first, and then archive the marker in a separate archiving run, so that the non-cumulative changes and the marker are stored separately in the archive.

The advantage of time slot archiving is that archiving runs can be carried out periodically using a single variant with a relative time restriction. The alternative archiving method described below does not have this advantage.

Archiving with selection characteristics

In some cases, archiving based only on time slots does not have the required effect of reducing the volume of data in the InfoCube. This is the case, for example, if the non-cumulative bearing objects are subject to high fluctuations, meaning that data is constantly being loaded for new objects and old objects are made obsolete. With time slot archiving, the end non-cumulative remains for all objects until all of the data in the InfoCube is finally archived, even if all the non-cumulative changes for an object have been archived already.

In this case, you want to archive and delete completely any objects that are no longer active in the InfoCube. This means archiving the marker as well. For other objects that are still active, you only need to archive the non-cumulative changes from the past, which means the marker is retained.

When you define the archiving properties for a non-cumulative InfoCube, do not therefore choose the time slot archiving method. Instead, select individually the characteristics that are relevant to restricting archiving runs. The time reference characteristic should also be included in this selection whenever possible.

Using this archiving method, for each archiving run, you need to create selection conditions in a variant. In the variant for the archiving run, in addition to the selection conditions, you can also specify for the selected characteristics whether or not non-cumulative changes and marker data is archived. If you choose to archive marker data, be aware that the time selection contains the maximum value for time.

Reloading data from the archive

As a general rule, SAP recommends that you do not reload archived data back into the InfoCube that it originally came from. It is better to reload the archived data into a structurally identical copy of the original InfoCube instead. For the purposes of analysis, both InfoCubes can be combined using a MultiProvider. This applies in particular to non-cumulative InfoCubes, but there are a few implications here as well.

The procedure for reloading archived data is different for non-cumulative changes and for marker data. Note the following:

·        Marker data and non-cumulative changes must always be reloaded from the archive in separate requests.

·        Marker data must be reloaded before the non-cumulative changes.

·        You are not able to reload marker data from the archive directly into the marker of a non-cumulative InfoCube. This is because the archived data is accessed by the export DataSource of the original InfoCube, which does not support non-cumulative initialization. We recommend the following temporary solution: You can use a full update to load the marker data as non-cumulative changes if you change the value of the time characteristic to a value that lies in the distant past (before the likely value of a lower threshold of the validity area) and if the request is compressed adjusted to the marker. The point at which the data is reloaded should be hidden by adjusting the validity area.

The validity area is not adjusted automatically if you delete data selectively (particularly in connection with an archiving run.) You need to use the transaction RSDV to adjust the validity area manually wherever necessary.

·        If the archiving run contains a mixture of non-cumulative changes and marker data, you can set a selection condition for the record type characteristic (0RECORDTP) during the extraction of the data from the archive into the InfoPackage to keep the non-cumulative changes and the marker data separate. 0RECORDTP = ’0’ indicates the non-cumulative changes and 0RECORDTP = ’1’ indicates the value of the marker that is stored as a non-cumulative change (= end non-cumulative.)

·        You must compress non-cumulative changes immediately after you have loaded them from the archive, without adjusting the marker.

See also:

Example of Archiving Data from Non-Cumulative InfoCubes