TM1 Forum

Hi,

I have a question surrounding the performance of TM1 cubes in comparison to memory size. In my simple brain my initial expectation of performance would be linked memory usage of the cube, i.e if it's very large it would naturally sufferer more in performance? Are am I being way too simplistic?

It's it possible to have a small cube, say less than 100MB and perform terribly through poor rules, cube dependencies or dimension sizes?

Size doesn't matter! (tee hee). It's not the size of the cube, it's the size of the view you are requesting. Also, how much does TM1 have to do as far as calculations is concerned, either consolidating or calculating formulas to get you there. Poor feeder structure, i.e., overfeeding, can hurt too because it can defeat TM1's sparse consolidation algorithm. Bottom line is poor design will kill the performance of almost any cube unless the cube is really tiny.