Fabric vs Databricks Day 3: Some Closing Thoughts
In this article, I shared some closing thoughts as we evaluate the best and the worst of Fabric.
The Good
1) Predictable pricing: Playing around through Fabric's offering knowing that I wouldn't break the bank (aka SunnyData’s account) unless I changed the capacity size was nice! This is the same peace of mind users have with their workloads, knowing the price is fixed.
2) Endorsements: I love the concept of endorsed and promoted objects (whether reports, notebooks, etc.). A while ago, I called this out as a feature that Databricks should seek to emulate on its tables from PBI. Very beneficial especially to smaller/mid-size orgs where some data management practices don't exist/are not possible.
3) Bundling of Tools: It is extremely convenient to have almost everything you need in one place , one of the things I enjoy about Databricks as well. Note: I do believe the quality of tools with Databricks is better, but Fabric's toolset is very complete.
The So-So
1) Power BI is fantastic, but I see little value to the end-users from the integration with Fabric. Little to no upside for PBI users. I didn't get to test Direct Lake, but from what I've read from very technical individuals, including those who benefit and who don't benefit from Fabric doing well, the benefits of Direct Lake are limited.
2) Fabric makes few claims, if any, about performance beating competitors, other than about PBI + Direct Lake. I speculate that it is because it is not faster dollar to dollar compared to others. I respect the lack of claims about it being faster than others. Testing would be fairly tricky due to a number of factors.
The Bad
1) I am in zero control of performance, other than perhaps timing workloads + Spark pools.
2) The UI is overwhelmingly busy, and when you know what you want, it is death by a thousand clicks. For a UI heavy product, this is completely a miss.
3) Organization is lacking. This alone is a reason why I would not recommend Fabric to any growing organization. Power BI was never known for being well organized and now, Fabric is built around Power BI. Let me put it into perspective: Schemas are in preview, and workspace FOLDERS, are in preview. Note: One saving grace here is the concept of task flows, which I think are a good.
4) Your business might grow at a rate of 10% annually, your Fabric cost will double when you reach capacity beyond bursting. I experienced this in the past with other Azure products, and it was painful! Would not recommend for this reason alone.
5) No Unity Catalog.
Closing Thoughts
+PBI has long been Microsoft's powerhouse and to a lesser degree, Data Factory. I believe it was a smart, cost-effective move by Microsoft to try to salvage years of work into other offerings like Synapse that didn't quite take off well on their own.
- The problem with the above though, is that there are many solid reasons why those other products didn't do as well on their own, but, here they are again!
Bottom-line: Avoid Fabric unless you are a small, non-tech org.
Helpful Links
You can find all of the articles in this series at:
Fabrics vs Databricks Conclusion
Also, would love for you to follow me on LinkedIn as well @ in/JosueBogran and Youtube @ JosueBogranChannel.