A Bit of Intro
If I recall correctly, I completed the first version of this data architecture diagram in 2012 when we used terms like "road map" and "blueprint" Back then, along with different terms, we were also using traditional SSIS, SSAS-MultiD and SSRS tools. Now we live in the world of cloud everything, although we are still driving from SRC-to-DST (source to destination). I'm up for whatever terminology you want to use, but can we agree that we are surely on a different highway? For my classical BI Blueprint, click here, but to see an Azure road map for BI, please take a look below.
Disclaimer: I create a different diagram for every engagement, so think of this as a suggestion, not a mold.
Azure Data Architecture BI Talking Points:
BI Advice from the University of Hard Knocks:
Conclusion of the Matter: I am not explaining every column in the data architecture because the columns in the above diagram are not applicable to everyone. For example, almost everyone needs a semantic layer, but not everyone needs a logical data store for operational reporting. Column #5 can be done in Spark as well as Data Bricks; instead of my telling you what the best solution is, let's talk about it. For every column there is a good, better and best solution, and good heavens (!) not everyone needs a thirteen point data architecture! All things in moderation, right?
I am asking, if you have taken the time to read this, please start planning before you start building! Opening Power BI and mashing up data from three different sources is generally not a scalable solution. Get started with a data architecture diagram and build a better BI house!
|Microsoft Data & AI||
Modeling for BI