TM1's order of aggregation across dimensions

[mce]

Hi,

I would like to ask a question for experts and would appreciate if I can get some useful replies.

What is the order of aggregation accross dimensions in TM1 cube that has many dimensions?
For example you may have the following dimensions in a cube in the following order:
- dim1
- dim2
- dim3
assuming that there is a top level consolidation in each of these dimensions and if there is no rules in the cube.
In calculating the cell that corresponds to the top level consolidation in each dimension, how TM1 calculates the final number? Does it follow a specific order for the dimensions in calculating the aggregations? such as do the calc in dim1, and then using the output do the calc in dim2 and then using the output do the calc in dim3?

Then the next question will be if there are some rules that will need to be used in calculating the final numbers, how this order of calculation is impacted?

[rozef]

See on IBM documentation for 9.5.x:

You have no control over the order in which TM1 performs dimension consolidations. Furthermore, depending on which consolidation path is optimal at any given moment, TM1 may alternate between paths. Consequently, you may request the Total, 1 Quarter value twice in the same session and get different results.

http://publib.boulder.ibm.com/infocente ... 916F3.html

Hope this can help.

[John Hammond]

You should be able to experiment as to what TM1 does in various circumstances by defining rules that override the consolidations in each dimension in the level below the top element. Under different circumstances the top cell will come out different.

I would suggesting testing what it does on a view consolidation to the top cell with and without zero rows and columns excluded