Hi,
we've added a big dimension (about 2G, 90-100 million rows) (it is actually a fact dimension) and a new measure group (based on the same table) into the cube to fine tune some MDX performance which contains certain measure, yes queries have that measure performance is much better now.
but I've seen some queries which do not really use the new measure group or the new dimension become slower (e.g. one of them 1.5 minutes to 2 minutes).
when I checked sql profiler "Query subcube Verbose" (EventSubclass is 22-Non-cache data), I found the duration for the event changed from 720 or so to 4500 or so. and there are 4 other events having similar increase doubled the time).
Should I accept the fact that it is natural the query will be slower even though it does not have the new dimension or measure group in the query, or should I start digging something fishy here?
Thank you.
-- Let us help each other to continuously improve.