Number Range Buffering can help cube loads if you your loads result in a good number of new Dim IDs being created with the load.
Without buffering, each row of data to be loaded that would require a new Dim ID to be assigned (because it is new combination of characteristics for that dimension) must go out and read the Number Range table for that specific Dimension and get the next available number. So if you add 50,000 rows to a cube that results in 5,000 new Dim IDs to be assigned, that means 5,000 queries that must go read the table. Now if you buffered 500 numbers at a time, that would mean only 10 queries to the number range table.
In a perfect BW world, someone in your shop would have the ability to go to your P system and make the change. Chances are, this ability is locked down and you will have to get the P opened to make the change.
This can also make a big improvement in the load time if you are doing a large initial load of millions of rows to a cube.
Also - just a reminder, line item dimensions don't benefit because there really is not a dmension table for them.
You must be a registered user to add a comment. If you've already registered, sign in. Otherwise, register and sign in.