Skip to end of metadata
Go to start of metadata

Adding characteristics to an InfoCube

Even when an InfoCube already contains  data, it is still possible to add new characteristics (e.g. with the technical name CAHR1). In such a case a new field ('SID_CHAR1') is inserted in the corresponding dimension table and all existing data records in the dimension table get the value 0 for this field. If you need certain values for the new column you must insert them with direct SQL. An alternative is to copy the cube, add the new characteristic and then load the data from the origianl cube into the copy. By doing this you can fill the field in a transformation.
See also SAP note 802697 Adding characteristics in dimensions of an InfoCube

If the charateristic CHAR1 is compounded (e.g. with the infoobject CHAR2) the situation becomes more complicated when CHAR2 is already in the cube. In general we in such a case SIDs for CHAR2 which are unequal to zero in the dimension of the cube. If now the new object CHAR1 is added we get a combination of SIDs in the dimension table which does not exist! Let's discuss this with the help of an simple example:

SID table CHAR1:

CHAR2

CHAR1

SID

 

 

0

v1

 

1

v2

 

2

v1

w1

3

v2

w2

4

Please note that the Data Types for the Characteristics are CHAR, meaning that the initial values (SID=0) are blank (' ') and not zero ( Data Type NUMC).

SID table CHAR2:

CHAR1

SID

 

0

v1

1

v2

2

Dimension (after CHAR1 was added):

DIMID

SID_CHAR2

SID_CHAR1

0

0

0

1

1

0

2

2

0

The combinations 1/0 and 2/0 for the SIDs of CHAR1 and CHAR2 do not exist! SID=0 for CHAR1 refers to SID=0 for the infoobject CHAR2 and hence the dimension is inconsistent! This leads to incorrect query results when e.g. when there is filter for infoobject CHAR2 and CHAR1 is in the drilldown.

Example of correct Dimension:

DIMID

SID_CHAR2

SID_CHAR1

0

0

0

1

1

1

2

2

2

These SID values (e.g.) would be correct, the values for the field CHAR1 (without the compounding) are always initial.
Since such inconsistencies cannot be repaired automatically you get the following warning when an infocube is changed this way: R7B 105. Please note that this warning is only issued as of release BW73x.
Please do not ignore this warning since this leads to wrong query results! The best way to repair such an inconsistency is to load the data into the InfoCube again. Then the SIDs are recalculated correctly based on the compounding.

  • No labels