Best way to develop and implement a solution

[dguevara]

Hi Everybody...

what is the best way to implement an enterprise solution working with large databases, because I can develop my model in Power BI Desktop and ok no problem, the problem is that there is still no solution like a on-premise server for Power BI and the cloud service is limited to 250MB so it is not scalable. then what would be the best way in this case?

Thanks for yours opinions...

[dguevara]

@Anonymous thanks for the comment, but I've never used that route so I have a question, what happens when Power BI connects to an SSAS model?
1- Power BI would act only as a component of visualization and cloud storage?
2- Can I use DAX into Power BI with that SSAS model or Power BI also load these measures of tabular model?

[greggyb]

When connecting to an SSAS data source in live explore mode, you are strictly consuming what exists in the SSAS model (regardless of Multidimensional vs Tabular). You cannot define new measures in Power BI when utilizing a live connection against SSAS. All of that modelling and measure definition must be completed in SSAS.

[Anonymous]

@dguevara For enterprise level solutions nothing beats going the Tabular model route. Rather than use the Power BI Desktop, if you develop your tabular model and deploy to an SSAS instance there are no limitations on size, and Power BI can direct query your SSAS instance using the Analysis Services Connector or Enterprise Gateway (which will replace the AS Connector eventually).

[AliSharifi]

@Anonymous, How about SSAS multi dimensional model, I guess it has been supported in the new Desktop realeae.

[Anonymous]

@AliSharifi Absolutely a viable option. I assumed based on the question that SSAS in general wasn't considered initially. Given that assumption, I recommended Tabular over Multi-dimensional as the learning curve is much lower initially, while the gains are immense. There are scenerio's for each solution in which one may be better than the other.

[bidgeir]

Hi

The 250MB limit is pr. dataset and due to great compression you can pack a lot into 250MB. In Power BI pro each user has 10GB of storage so you can potentially have 40 250MB datasets pr. user. 

The main way to keep down data size is to be careful when you import data in such way that you only import the data you need. This is a columnar store database so number of columns and their sparsity is very important. Don't for example import id columns on tables if you don't need them as they are unique for each row. 

Hope that helps you. I'm sure others can pitch in with their experience as well