The case for Power BI Premium?
Some time ago a colleague of mine - let's call him John - was bouncing up and down, saying we need Power BI Premium. A report was not performing very well and he had read that Premium would give us dedicated capacity. A quick glance at the problem made me realize he was having trouble with that report in Power BI desktop. A brief smile appeared on my face.
"John, you already have dedicated capacity", all that 8 gig of ram and Intel I5 processor are all yours". He did not realize that a Power BI report in Power BI desktop consumes local resources and does not make use of the cloud. At least, not for rendering anyway. If there is a direct query or a live connection, the data calculation is off loaded to the data source. Which also means that if that data source is not Power BI, you also won't leverage www.powerbi.com to do the heavy lifting for you.
In this case, we were on a department that uses SQL datawarehouses and Analysis Services models to serve Power BI reports. That means that the only workload www.powerbi.com is doing for you is showing the calculations, not actually performing them.
When he understood that, we both agreed Premium was not yet needed. We did not have thousands of users, nor did we need the added capacity. Keep in mind this is some time ago and the premium offer has changed over the years.
So if you are wondering if premium is something for you, let's think about that together. If you are using reports that have direct query connections or live connections to Analysis Services, you need to realize that you will not get any performance gains from going to premium. Unless you want to swap your Analysis Services models for Power BI models.
So when could it be for you?
I like to start of by listing a few important considerations:
If you have a lot of reports with import mode - which moves the workload to Power BI - then it could be interesting if you need to scale up. Premium data sets are up to 10 gigabyte, without premium the max is 1 gigabyte.
Larger self service bi models could benefit from this. Often IT delivers reports through datawarehouses and Analysis Services cubes, those reports have their data model externally via live connection or direct query. Unless you plan to move the data model to Power BI they will not benefit.
Power BI requires a pro license to read / view shared content. If you have hundreds to thousands of users, it may just be cheaper to go with premium, which allows free users to read shared content. Go to https://powerbi.microsoft.com/en-us/pricing/ to check that out for yourself.
If you have lots of SSRS reports which you do not want to or cannot convert, with premium you can host them in the cloud. This makes it possible to have 1 reporting environment. You can publish your RDL files and have them co-exists with your Power BI reports.
Incremental data refresh
With premium, you can refresh your data sets incrementally. This makes it possible to only refresh the current year for example.
With premium you get a lot more options for using AI: text analytics, vision and using Azure machine learning models. However, if you truly require custom build advanced AI, you're better off with Azure Data Bricks and create the models yourself.
The bigger picture
The real question is; how do you want to process your data? Premium does not replace your Azure Datawarehouse, your AI platform or your enterprise scale data processing. What it really seems to be for, is to further empower the self service BI user. To close the gap between IT managed BI and self serviced BI.
But if that were truly the case, why is there talk of Power BI premium versus Azure Analysis Services? The latter obviously is not a self service component, but for IT to serve a single data model to lots of reports and scale out if needed. Let's take a look at what Microsoft are saying:
Power BI Premium is the focus for enterprise BI and the primary target for future investments. In time, Power BI Premium will provide a superset of the capabilities when compared to Azure Analysis Services.The clear future direction is Power BI Premium.
And here may be the best reason of all to consider premium: Microsoft is planning to make Power BI premium the superior tool of choice. With scale out coming around mid 2020, the decision will be more easy to make.
So, does it replace your data warehouse? No, it certainly will not. But it might very well replace your Analysis Service models. Now if they make Power BI files easy to track in source control and make it easy to generate them, IT departments might find themselves looking at premium too.