Fabric IQ: Another semantic layer - Time to Market

I have been playing with Fabric IQ (a different workload in Fabric), Ontology, and Graph, and while I like the way this is set up for the future, isn't this one more direction for IT to change towards?

What and Why

What: A semantic layer that defines business entities, relationships, rules, and context independently of physical data structures. This ties to the data in lakehouses, eventhouses, etc. Now, this can be displayed as a graph because of relationships between entity types (loosely tables). Why: This setup of ontology is where it excels in displaying the data as relationships between entities in a graphical format. You don't have to struggle to identify what "Sales" is anymore because the definition is defined by you.

Key Lessons

1. How to get there

Check this website to make sure your region supports Fabric IQ: Availability. Our tenant was in South Central, so I didn't see an option for Ontology. We created a new capacity in a different region that supports it, but still didn't see the option. I was able to generate Ontology using APIs though. This is a lot of manual work if you have to define the entities and relationships.

2. Time it takes to generate everything

We have what I would say is a very small semantic model, but the generation of it took a lot of time. It could be capacity constraints too. We were on F4 and then bumped up to F16 but still saw the same time.

3. Entities, Binding and Properties

Just like a semantic model, much care needs to be thought out before starting this. If you have a semantic model with measures and everything, that is a good start. If you don't have it, then it is quite an exercise. Plan and prepare. If you have a lot of tables, one wrong relationship will set you back. Defining these relations is key. Example: If you have sales and vendors, define the relation as vendor "has" sales.

Conclusion

This is a good concept if you have time. If you're just converting all your reports into Fabric medallion layers, then you might have to think about it again as to where to spend the time (keep converting or think about IQ). I understand that everything is moving at a fast pace these days, but MS needs to understand that not all companies are equipped to move at the same pace (MS is not forcing). I also didn't try the operational agent yet, which I didn't see in our environment. I will publish once I get my hands on it.

Additional Resources

Ontology

Ontology Tutorial

Questions? Drop a comment below.