Hi Tina,
For my allocation models I typically use an export and import approach for a few reasons:
- Reading and writing to the same cube can be slow where the writing will cause the cache to be cleared and thus subsequent reads to be slow
- Exporting the ratios/proportions to files allows the same file to be reused if necessary
- Simplifies the process in my mind by separating the steps
- TI spread function was just too slow to be practical
Essentially then you would have a caller/wrapper process that does the following:
- Calls a child process which exports the ratios to one or more files depending on your model, number of drivers etc.
- Builds a view based on the values to be allocated
- Calls a child process passing the context of the parent cell to be allocated and the amount to be allocated
- Child process reads the relevant ratio file and for each record takes the ratio, multiplies the amount to be allocated and then writes to the cube based on the member the ratio relates to.
Hope that helps.