My dataset have 1 Fact table with around 1000 columns and 1000 dimensions, for each dimension, i need non null count, percentage, and mean.
What will be the most effecitve design? 1 cube or many cubes or 1 measure group or many measure groups, will partitions help ?
Currently i created 1 cube, 1 measuregroup when output all, the quesry runs for ever.
Can I put count in 1 mearure group, percentage in 2nd and mean in 3rd measure group ?
View Complete Post