This page has 0 threads | Add post

Blogs

Finding our blogs useful? Please consider helping us to share more!

Using partitions/processing to load a large fact table into a tabular data model
This blog shows how you can load the empty structure of a large fact table into an Analysis Services tabular data model, then use partitions to bring in the data bit by bit.

Posted by Andy Brown on 23 July 2018

In this blog

Importing the table structure
Creating partitions on the table
Processing partitions

How to load a large fact table into SSAS Tabular

So you have a large fact table, consisting of perhaps millions of records:

The sample fact table we'll use for this example. You can try out the steps below yourself by downloading the Excel workbook containing these rows.

You want to load this into a tabular data model, bit by bit. How to manage the process?

I couldn't find an official approach anywhere, but the method shown below certainly works. I'm using Visual Studio 2017 and SSAS Tabular 2016, but think this method will work for all software versions.

Importing the table structure

The first thing to do is to import the table's structure:

While importing your fact table, click on the Preview & Filter button.

Set a filter which effectively excludes all of your rows from import:

Here we set the filter to import only rows where the purchase id is less than or equal to 0 (which can never happen, since this is an identity column).

The result? An empty table: