Last up in this month series is a review of the preview features we're seeing drop into both Fabric and Power BI.
As always, do remember that it is not recommended to use preview features in production unless absolutely necessary. By using them in production you are carrying the risk of breaking changes, and without SLA level support.
First up, Power BI.
Power BI
Visual calculation updates
This month we've seen:
- Parameter pickers for required parameters when a function template is used.
- Coloured highlights similar to the way that Excel works.
- Change in behaviour when using columns in a function so that a value instead of an error is returned.
- Added visual calculations to the explore view.
For me, the largest update in this change is adding the visual calculation option to the explore view. This feature will make it easier to allow train of thought analysis to move into areas not natively supported in the semantic model.
Whilst the coloured highlights are welcome, my concern is that this will become a bit misleading for new users. I can see those users, who don't understand about the row and filter contexts, to assume the Power BI works exactly the same as Excel and get the wrong results.
Org app updates
This month we've had a couple of new features for org apps:
- Paginated reports are now supported
- Copilot is now available for consumers
Power Query editing in the service
This month they've added the ability to edit import models in the service. This addition means that you now have an end to end authoring experience in the service, making Power Bi dev on non-windows devices an option.
Whilst it's not feature parity quite yet, it's definitely a big step forwards.
Fabric
Data engineering
Variable library integration with Notebooks
This one is a very welcome addition. By adding variable library support, it means that we can store Lakehouse id's, names, etc in a central location that is pre-configured for dev, test, and production.
The key thing to remember is that these strings are stored as plain text in your source control. That means that you must not store any sensitive information into a variable library - for those you need to use key vault and integrate with key vault still.
Inline code completion
Driven by Copilot, the inline code completion for Python is now available in Fabric Notebooks.
T-SQL and Python notebooks against Datawarehouse
Materialized Lake Views
This one means that you can use Materialised views in a Lakehouse to simplify medallion architecture. The system automatically manages the update to ensure that these views are refreshed when underlying source data refreshes.
Personally, I'm not a fan of anything that obfuscates code away (yes I'm looking at you SQL server triggers), but given these are included in the data lineage at least means they aren't completely obfuscated. The one thing I do like about them is the out the box alerting for data quality conditions.
Definitely one to keep an eye on.
Data Warehouse
Result set caching
The only real update this month, and one that should have come pretty close after the original release. But it's here now, and definitely welcome.
That's it for this month, and probably until the end of September. Have a great summer and catch you all on the other side!
Comments
Post a Comment