We have some very mature cubes built from SQL 2000 days and have been migrated up to 2008 R2 at this point.
They have snowballed to become quite complex. Their data volumes are very small, the biggest one is 200 MB, the smallest is 4 MB.
But their MDX Scripts are quite complex (5,000 - 13,000 lines+), involving Solve_Order extensively.
One approach I am thinking of is to cull deprecated MDX from it, reorganize the MDX calculated members and rewrite as much of it as I can using SCOPE() assignments and banishing Solve Order as much as possible.
Have others tread down this road? Does this approach seem sound?
Getting rid of Solve Order in this way makes sense to me, but have never done it before.
Microsoft is conducting an online survey to understand your opinion of the Technet Web site. If you choose to participate, the online survey will be presented to you when you leave the Technet Web site.