November 20, 2009 at 7:59 am
I've got a cube 🙂
Cube has application performance data. Basically as screen load times.
Application load times can be zero when the underlying systems determine that the locally cached data is up to date enough.
Two different groups.
Use Experience group, wants MIN/MAX/Average/and count of uses for all data.
Development Group: Wants MIN/MAX/Average/and count of uses for data where the loadtime is non-zero.
I managed to solve the Average and count values by adding a calculated field onto my measures table that is either a 1 or a 0 depending on if the individual loadtime is 0 or not. Then I just sum that column and divide the aggregated total execution time by the summed result for the average execution time for non-zero execution times.
But I'm at a loss how I can modify the MIN/MAX functions built into the cube designer to have it take the Minimum execution time that is not Zero.
Anybody have an idea?
November 24, 2009 at 10:58 am
You may have to add another calculated column that holds the values for the non-zero rows. Use NULL for the zero rows so that you can use the COUNT aggregation to eliminate the NULLS for your average and COUNT.
There may be an easier way, but without any other suggestions, I'd go the calculated column route for this one too.
November 24, 2009 at 12:08 pm
Actually came up with something that worked quite well.
I created a view off of the measures table that just excluded the 0 values. And then brought that in as a second measures group. Performance wise it's not causing any issues with the cube and I have the benefit of two seperate measures groups... one with 0 values, and one without to make it easier for the users to use. 🙂
Viewing 3 posts - 1 through 2 (of 2 total)
You must be logged in to reply to this topic. Login to reply