Playback speed
×
Share post
Share post at current time
0:00
/
0:00

I recently wrote 3 articles about Fabric vs Databricks, focusing on Fabric from a neutral standpoint, but don't confuse my neutrality as saying they are equally as good. For most businesses and organizations, Databricks is the right choice. Here's why:

Cost

With Databricks, you pay only for the compute that you use, when you use it, period. With Fabric, you are ALWAYS paying. It is the anti-cloud. Use it or lose it. You can borrow from the compute ahead and throttle yourself, but you can never get back what you paid for and didn't use.

Performance

I never trust benchmarks as to which platform is faster than the other (read: Databricks vs Snowflake benchmarks), but benchmarks do help provide some directionally correct guidance. When Fabric is absent from performance benchmarks and there is little marketing as to its performance vs competitors, that is a big tell-tell.

Governance

Unity Catalog is arguably one of the top 2 reasons why customers use Databricks. It makes data governance simple. Fantastic lineage capabilities, reporting through system tables, easy access control management, the list goes on. Meanwhile, Fabric users have the lighter version of Purview, which does offer some unique that UC does not have, but lacks many of them, and is more data observability than governance, an area that Databricks does a fair job in as well.

Orchestration

Databricks Workflows work very well, are feature rich, and arguably are of the most mature Databricks offerings, second to UC. Solid at retaining job execution history, orchestrating pipelines efficiently, and allowing compute choice flexibility. Also job compute in Databricks is pretty darn cheap. Fabric offers Data Factory, which is okay for basic orchestration, but anything beyond basic orchestration, and it gets painful.

ETL

Databricks doesn't offer good low-code solutions out of the box (an area Fabric does well in with Dataflows), but it is far more powerful when using code than Fabric can be with or without code. Databricks gives you plenty of methods to do ETL, both vendor and non-vendor specific. If no code/low code is your thing, Lakeflows (coming in the future), as well as 3rd party tools such as Prophecy can help bridge this gap.

Dashboards

Power BI is the market leader, no doubt about that, and why Fabric bundled others services around it. I personally love using Power BI as well as Sigma for dashboards. That said, Databricks works extremely well with Power BI, even allowing you to create starter models straight from Databricks' UI.

Coding Experience

The SQL Editor and Notebooks from Databricks offer great coding and data exploration experiences. Surprisingly, even leveraging charts in Databricks is easier than in Fabric's equivalents.

ML/AI

Databricks is THE platform to emulate in this category.

Closing Thoughts

Fabric is good for small businesses that have a very limited tech skillset, but for most other businesses, Databricks is the top choice, while Fabric is in 4th place.

Helpful Links

You can find all of the articles in this series at:

Fabric vs Databricks Day 1

Fabric vs Databricks Day 2

Fabric vs Databricks Day 3

Fabrics vs Databricks Conclusion

Also, would love for you to follow me on LinkedIn as well @ in/JosueBogran and Youtube @ JosueBogranChannel.

Josue’s Substack
Josue’s Substack
Authors
Josue Antonio Bogran Micher