Data Sources vs Performance Calculation

[timsleeper]

I have a report with a source not that big (~300-400k rows) but it relies on some pretty heavy calculations (http://community.powerbi.com/t5/Desktop/Pareto-80-20-calculation-taking-too-long/m-p/24255).

Up until last week, I had the report built with Power BI desktop with on-premisses data sources and then published to the cloud with the desktop app. Performance was sluggish at best. Then I made a little experiment uploading the data to a Azure SQL Database, set a copy of the report pulling the data from Azure SQL via DirectQuery and then publishing the same way. The performance was way better.

My questions is, does moving the data to the cloud and using it as a DirectQuery source improves the general performance of calculations? I thought that all calculations were made on Power BI but if the Power BI + DirectQuery + Azure Cloud combination improves performance, I'll definitely consider hosting some of the data on the cloud.

Cheers,

Tim

[ankitpatira]

@timsleeper It could be more about data retrieval from Azure SQL DB as it stores data across multiple nodes, data retrieval would be better than your on premise data source. 

When you had on premise data source with power bi desktop did you use DirectQuery or import then? As I would assume you might experience performance delays in calculations if you were using DirectQuery from powerbi.com to your on-prem data source and then doing calculations on it.