Attribut change run

Question

HI,
I've one doubt about ACR.Since we deactivate all the aggregates and activate all the aggregates after attribute change run to speed up attribute change run whenevr we undergo system master data mass update?
Is this the correct way to handle this and another My question is if we are not loading any data in some xy cube but whenever we are doing mass update to master data we still deactivate that xy cubes aggregates and then activate the same.
Now if we are not laoding any data in the cubes then why we need to bother about aggregates of that particluar cube.I think we don't need to touch aggregates of that xy cube because we didn't loaded any new transaction data in it(Note cube already has previous years data ,I'm reffreing new data is not being loaded into cubes but this cube still being used in queries)
is this something related to historical records?
Can anybody clarifies my doubt?

Answer 1

Hi Rubane,
    Its like this, an ACR is used for updating master data changes. I'll try and explain with an example.
You have material in the cube and an attribute material group of material is switched on a navigational attribute.
Now since this is a navigational attribute, you can add it to an aggregate to improve your query performance.
Now a material group can be changed for a material at any given point of time.
say material X earlier had material group A but now has material group B.
In youe aggregate, you  have all values corresponding to material X still assigned to material group A
eg aggregate Z
Material Group  material key figure
A                      x            100
This is where an ACR comes into picture.
If you run an ACR for material, it activates the master data and then realligns the aggregate with any changes
So now your aggregate looks like this
Material Group  material key figure
B                         x            100
So now to answer your question, yes we need to realign the aggregates even if the cubes are not being loaded. This is to take care of any master data changes that occur.
As for your second question, i do agree that its better to have aggregates with a '-' rather than none at all, since these help in the selection. Rather than selecting data from the cube, the query selects the data from the aggregate which is faster.
However the draw back of such aggregates if that they take a long time to fill. and nothing really changes i.e. very few records are summarized. I would suggest in your case, that you look at the aggregates and check the usage field. If you are continously activating and deactivating the aggregates, this really wont give you any idea. However you can still check this field. The more the value, the more times the aggregates is used.
Generally there are aggregates which have no usage at all. These can  be deactivated which will save you some time.
Also check for the aggregates which are taking time. You can check the logs for the change run. You will  see some values like /BIC/F**** where the ****** indicates the aggregate no. Go to table RSDDAGGRDIR and find the aggrcube field and enter this value. You will find the aggregate. You can then run some checks in RSRV for the cube. You can also try deactivating and then reactivating the aggregates. This will remove any inconsistency in the database for that aggregate which will save you time.
Regards.