rigid vs. flexible when doing full process

Given the assumption that the cube is small enough that we'll always do a full process on the cube...

Is it fair to say that there will not be a difference in MDX query performance depending upon whether all attribute relationships are rigid vs. flexible?

I know flexible vs. rigid impacts performance when incrementally processing dimensions. And I know it impacts aggs... as I understand it aggs involving flexible relationships are stored separately from aggs involving rigid relationships. I'm trying to validate whether aggs are stored less efficiently because of flexible relationships... and exactly what impact on query performance this will have.

Hello. The performance guide pointed to at the top in this discussion group will help you answere that question.

Regards

Thomas Ivarsson

|||

Thomas. Thanks for the reply. However, I've read the perf guide cover to cover, and I don't believe it answers my question about whether an MDX query off a flexible agg versus a rigid agg will perform differently. If there's a sentence I missed, please let me know.

So I'm considering this thread still open.

|||I don't beleive there is any difference in MDX performance w.r.t. rigid vs. flexible relationships.|||

I agree that there is nothing about the difference in query performance in the Performance Guide between flexible and rigid aggregations

And now we know from the primary source that there is no difference in this respect.

I have heard a different story in a Blog entry by someone else than Mosha. That was wrong.

Very good question!

A question to Mosha. Why should we use rigid attribute relations?

Regards

Thomas Ivarsson

|||

A question to Mosha. Why should we use rigid attribute relations?

If you are always going to do full process of the cube, then I don't see the difference between rigid or flexible relationships.

|||

Thank's for the reply Mosha.

I might be wrong but I think the question was about query performance irrespective of processing performance.

Lets say that you have a theoretical cube that you load only once and never update. What about the query performance between flexible and rigid attribute relations? I know that it is very hard to say but if we have one cube with only flexible attribute relations and one with rigid, what cube will perform best?

We also assume only natural hierarchies in both cubes.

We assume the same UDM in both cubes.

Kind Regards

Thomas Ivarsson

|||My answer about MDX query performance was that I don't beleive there is any difference in this scenario.