Better database option for bigger data for PowerBI

[JustinDoh1]

I am trying to get a better understanding which database would work best for Power BI Service and Power BI Desktop for large data.

Reporting group at my company has been developing Power BI reports for last four years, and we only have been using Azure SQL database for our backend data.

As our data gets bigger and more line of business request for more complex reports (asking us to create bigger data warehouse), I am trying to understand if there is better architecture to pull data (than using Azure SQL).

We have been using Azure so, trying other DB (beside Azure SQL) is not an issue, but it would take our time to develop any pipelines etc.

Recently, one of our Azure SQL databases had to increase DTU (like compute power) to almost 300, and it is consuming a lot of money per month (about $400/month).

I know Azure SQL is not all ideal for analytical data, but it is good for transactional data, so I am curious whether exploring other db option (possibly Cosmos DB) is a better approach. 

Please let me know.

Thanks.

[v-sathmakuri · Accepted solution]

Hi @JustinDoh1 ,

Thank you for reaching out to Microsoft Fabric Community.

Thank you @Deku  for the prompt response.

As mentioned by @Deku , you can use data warehouse and lakehouse in Microsoft Fabrics.

You can start with F2 or F4 capacities, and auto-pause and resume features save money during low activity.

Migrate reporting datasets to Fabric Warehouse or Lakehouse.

I recommend reviewing the documentation below:

https://learn.microsoft.com/en-us/fabric/data-warehouse/ 

https://learn.microsoft.com/en-us/fabric/data-engineering/lakehouse-overview 

https://learn.microsoft.com/en-us/fabric/data-warehouse/data-warehousing 

If this post helps, then please consider Accepting as solution to help the other members find it more quickly, don't forget to give a "Kudos" – I’d truly appreciate it! 

Thank you!!

[v-sathmakuri · Accepted solution]

Hi @JustinDoh1 ,

Thank you for reaching out to Microsoft Fabric Community.

Thank you @Deku  for the prompt response.

As mentioned by @Deku , you can use data warehouse and lakehouse in Microsoft Fabrics.

You can start with F2 or F4 capacities, and auto-pause and resume features save money during low activity.

Migrate reporting datasets to Fabric Warehouse or Lakehouse.

I recommend reviewing the documentation below:

https://learn.microsoft.com/en-us/fabric/data-warehouse/ 

https://learn.microsoft.com/en-us/fabric/data-engineering/lakehouse-overview 

https://learn.microsoft.com/en-us/fabric/data-warehouse/data-warehousing 

If this post helps, then please consider Accepting as solution to help the other members find it more quickly, don't forget to give a "Kudos" – I’d truly appreciate it! 

Thank you!!

[JustinDoh1]

@v-sathmakuri Thank you for your reply. I am still trying to learn what other options are there.

I hear "Fabric" all the time, and I am not sure how our group can get started if needed.

Don't we have to pay extra money to get started?

Ideally, start up cost and other set up costs should make sense with better performance. 

What is most common way to do this?  Fabric Warehouse or Lakehouse?

Sorry, I am so new to this.

[v-sathmakuri · Accepted solution]

Hi @JustinDoh1 ,

Microsoft typically offers a free trial that provides access to the services, including features for data storage, processing, and analytics. The free trial gives you a certain amount of capacity, allowing you to explore and test out various features without incurring any costs upfront

You can check the fabric capacity pricing below . You can start with F2 capacity for doing more analysis if required.

https://azure.microsoft.com/en-us/pricing/details/microsoft-fabric/ 

A data warehouse is ideal for structured, cleaned data where fast queries and analytics are needed.
A data lakehouse is better when you want to mix raw and structured data for large-scale analytics.

If this post helps, then please consider Accepting as solution to help the other members find it more quickly, don't forget to give a "Kudos" – I’d truly appreciate it! 

Thank you!!

[Deku]

Why do you need a transactional database, a warehouse or lakehouse would be preferred for analytical workloads. Something like fabric or databricks

Cosmos would not be well suited, I doubt you need the redundancy, high availability etc conferred by it