SSAS Tabular: Calculating the actual compression rate

We always hear about the power and compression ratio of tabular databases and Power Pivot, and how the Vertipaq compression engine can compress things up to 90% of what their source is. But, this opens up a lot of questions. For example, how can we properly size our tabular hardware solution in the first place? As in, how do we ensure that our hardware we’re setting aside is the proper amount. Unfortunately, just saying “Give me all the memory you can”, isn’t going to make our virtualization/networking guys a) love us, or b) happy. There are generally two steps in the iterative process for sizing:

Step 1: The initial estimate

Step 2: Refining the estimate and comparing the actual

Step 1 is relatively straight-forward. The general size of of a tabular database can be computed by running a simple calculation against the source database. First, run sp_spaceused against the database…