Відео

Deployment pipelines are not fit for purpose for pro dev. No dataflow support, and no parameterisation for Data Pipelines to name a few limitations. We use them for Semantic Models and Reports only for which they're fine. Everything else is Github and Actions.

Writing in the main chat box like you are doing there, that doesn't actually train the model. If it did, the custom instructions that would accumulate behind the scenes would be an atrocious muddle of unintended and countermanded orders as you trial-and-errored your way through it. Also, it literally says there, "conversation history is off" - no context retained. You need to copy your training instruction discoveries into the custom instructions "Notes for model" box on the right for the model to retain the lessons

The dark mask of Databricks

Datapoint annotations are going to be awesome. I'm also hoping we can eventually create our own prompts for data insights. Maybe even regenerate a response when the report refreshes.

I agree with the broad strokes here. I will say that when it comes to consumption and performance, choosing the right data store and data movement tools becomes more important. Dataflows are incredibly intuitive but can be 2-5x more expensive than pipelines or notebooks, from what I've heard. Microsoft could be doing more to outline the golden path for small businesses (<100 employees). I think the all Pro license orgs are feeling left behind right now.

It’s supposed to be easy.

Data load from Lakehouse files to tables in SQL endpoint failed due to different delimiter formats with CSV files. I modified your data load Notebook to take into account the delimiter and got it working for me. e.g. for all files with pipe delimiters add the option("delimiter", "|") #df = spark.read.option("header", "true").format("csv").load("Files/AdventureWorks/DimAccount.csv") df_pipe = spark.read.option("header", "true").option("delimiter", "|").format("csv").load("Files/AdventureWorks/DimAccount.csv") df_pipe.write.mode("overwrite").format("delta").save("Tables/DimAccount") comma delimiters worked by default but you could define those in the corresponding way with option("delimiter", ","): #df = spark.read.option("header", "true").format("csv").load("Files/AdventureWorks/DimCurrency.csv") df_comma = spark.read.option("header", "true").option("delimiter", ",").format("csv").load("Files/AdventureWorks/DimCurrency.csv") df_comma.write.mode("overwrite").format("delta").save("Tables/DimCurrency")

Hey. Interesting point about scheduled batch jobs vs reacting to storage events. Recently I watched an early Databricks talk on Delta Lake (4 years ago). They mentioned that one of their first big customers to try structured streaming (replacing batch) was able to use far far less servers due to increased efficiency and spreading the load over 24hrs. For those of us now working with Fabric, with a reserved capacity, spreading the load over 24hrs makes a lots of sense. It might also save you from upgrading. But it does introduce a new problem. One that Airflow can assist with. That being: too much getting started at the same time. I.e you no longer have a fixed pattern of load.

Do you maybe know what the SQL query variation which is set to 3 by default, is supposed to do?

Hi Chris! I don’t understand the SMB strategy… because if you pause your capacity wouldn’t end users lose access to the Power BI reports? Are the semantic models and reports located outside of the capacity?

Thank you so much for this!!!!!!

thank you very much, love from brazil and china!!!!

walking through the content already published is not all that help. I want to see it working and in action.

Please attach link as well for ref. And we are still below Tableau price ....

Thanks much

All the Fabric improvements need to be paid.

I was more concerned when MS dropped the Power BI Premium licensing like a P1 (P1 = F64). As a nonprofit we got tremendous nonprofit discounts on the Premium Licenses like the P1. There are no nonprofit discounts on the F skus as they are on the Azure side. There are some generic Azure credits, but we were already using them. I was really hoping MS would offer more help on he F sku area.

Hoping you can help with an ongoing issue I have. I have a powerbi dashboard i built using regular accent to fetch some SharePoint lists and uploaded it to PowerBi. Now in PowerBi I want to change the credential from my account to an SPN. I've read what feels lie a thousand article describing the process 99% all the same yet when I edit the permission select SPN put in the tenant ID the app id and the secret it fails 100% of the time. It's got Graph sites.read.all, user.read and SharePoint Sites.Read.All api permissions configured. All granted. Everything seems right but gives me the error failed to retrieve oauth token 100% of the time. Am i missing something else? More API permissions maybe? Do i still need to create a group, add the SPN to the group and assign it access the Sharepoint site in question?

Ever since I was a kid, I always wanted the highest res / best quality. Sometimes I cheat and screenshot icons. But I also have to double-check I didn’t grab the old version. This is nice. Thanks for the tip and resources.

Thank you for putting all of this together and sharing!

2957 icons, in place in architecture diagram

Hi! This helped a lot to update the semantic model if I make changes to the Excel. It appears almost immediately. But If I make a report from the semantic model, for some reason that isn't updated. Do you have a solution?

Chris: That’s a good point you made about queries across workspaces, and how it relates to the capacities that power them. I think about this all the time. Especially when I think about the potential for other departments to consume some of the capacity in a workspace controlled by my team. For the moment I’m recommending different departments add a table via shortcut, then run sql via the sql endpoint in their own workspace (and hence their capacity). I don’t know how this arrangement would interact with the new Fabric SQL databases. I suspect, to your point, it would work as the same capacity would be doing all the work.

Hi, thank you for the tutorial, was super helpful! However I have a question, for some reason when I make updates to the excel file but ensure the name is the same and hit refresh on powerbi, the changes are not reflected. My goal is to upload to powerbi service and I will be updating this file. What I understand is that through this method a cach file is being created which does not get updated. To see the changes I had to change the file name and then uplaod the data source again. Thank for your help!!

You had me at “history”

why not to use workbook or excel options or in other words when to use them over web?

Great stuff. Keep it up!

thanks for the 'resume' tip!

🎉Stay out of trouble

As much as I find Miguel likable and support his work, it was at times uncomfortable to watch him struggle with the visuals. I hope he doesn't lack motivation and that he'll provide us, the end users, with a much better product.

One of the few episodes I've watched more than once. Great stuff. Thanks for this!

GREAT

Would love a part 2 into the mind of Miguel 🧠

Great video! I wonder if there any way to directly access the quality metrics to use this information as part of data observability report or just do quality checks as part of a pipeline before ingestion

If he's struggling so much, how can we not?

Two of my faves! Thanks for the Christmas present to all of us!

Nice! You can tell when someone loves data! 😎👍👍

Could the GitHub repository mentioned be added? Great video!

This is very insightful. Thanks for sharing.

This is fantastic u r brilliant..tks

ideas.fabric.microsoft.com/ideas/user-profile/?user=d8bbad6d-747e-ef11-ac20-7c1e521a63a7

@KratosBI here is the idea submitted to Fabric team ideas.fabric.microsoft.com/ideas/idea/?ideaid=9f420002-d2be-ef11-95f6-000d3a7a93ec

Its item level security right?

I was in an interview with one company recently for Fabric admin profile and they ask me how to import bulk workspace id alongside report id and other things using service principal authentication and power bi rest API admin tokens.. Can u pls make a video on such bulk user admin downloads using same...

Do you know of any way we can use this outside the Fabric Notebooks? Like in a separate hosted chatbot or app?

So to clarify for sure - the Org App is now an object within a workspace, and doesn't act as some separate, managed, hidden workspace any more?

❤😂😂

will it work if we are using shared dataset model? Multiple reports connected to single dataset.

Awesome! You are not the Kratos. You are the Zeus of BI😀

Thumbnail needs the health bar from fighting games.