Hi!
Ive got strange kind of excessive readings on my cube.
Cube is apprx 150gb with 20+mgps, 20+ dims, 100+CalcMeasures, so ill give a simple example.
Slice properties are filled (by months).
Lets say the cube has
2 physical measures: sales and stocks.
2 dimensions: periods and regions
So i run simple queries:
1) period on rows, sales on columns. No filters.
It runs very fast (5 secs )
2) period on rows, sales on columns. Filter: multiselected regions (2 regions of 100 excluded).
It runs very slow (4minutes).
And despite of i select SALES MG in query, in the trace i can see SSAS reading all the partitions of STOCKS measure group.
3) Commented all Calculations in cube.
running the same as 2 (period on rows, sales on columns. Filter: multiselected regions (2 regions of 100 excluded).)
It runs very fast (5secs)
So the problem is not even in using "bad" calc measure, but in the fact "bad" calc measure exists.
Definetly ill find "bad" script part uncommenting all calcs one by one, but may be you can point me with advice?
UPDATE:
Found that "bad" part of script are Dynamic Sets definitions which are using second Measure Group STOCKS. So even if this MG is not mentioned in query ssas reads it to create dynamic sets. This fact slows down any cube browsing.
So now the question is: i use dynamic sets to avoid problems with totals and multiselects. And they slow down browsing of any measure a lot. What can i do to avoid this slowing while quering measures that are not linked to this dynamic sets?