PBI Service Dataflow: Import Model same model Deployed to AAS?

[pawlowski6132]

I developed a Tabular Model using Visual Studio and deployed it to Azure Analysis Services. Since it's virtually impossible to schedule a refresh of that model in Azure, I was hoping to use the Import Model function within PBI Service Dataflow creation process.

I can't figure how which JSON file, or how to create one, from my existing Tabular Model I created.

Is this even possible?

thanks team in advance.

jrp

[v-shex-msft]

HI @pawlowski6132,

Model.json file is exported from a created dataflow entity, I don't think you can directly get this file from Azure analysis service tabular model, they are different things which stored with different types.

If you mean get data from azure analysis service, you can try to create this connection on power bi desktop side, then paste this M query on 'dataflow-> blank query' to use Azure AS Datasource as dataflow entity:

Connect to data sources for Power BI dataflows#connecting-to-additional-data-sources 

Regards,

Xiaoxin Sheng

[pawlowski6132]

Thanks for your reply @Xiaoxin Sheng. Yes, that's exactly what I was trying to do; export the equivalent of the Model.json file from my AAS Tabular model to import into Dataflow. Scheduling refresh of AAS model isn't practical.

Further, not sure If I should even thing about dataflow because I'm failing to see the value the more I look into it. If I can already connect to datasource and crate modeling in dataset (which I can share on workspace) why would I ever need a dataflow??

[v-shex-msft]

HI @pawlowski6132,

I think you still not understanding my descriptions.

Normally model.json is exported from existed dataflow entity, it is hard to manually transform from specific Datasource.

For manual operations, you need to refer to metadata format and find out all properties and correspond contents id on the azure side(table id, table field name type and id, root source content path) then copy these to model.json file. (notice: you can't confirm if these properties can be recognized by power bi correctly)

Attaching Manually-Created Common Data Model Folders In Power BI Dataflows 

The metadata file (model.json) for the Common Data Model 

For these complex steps, I think use blank query to create a dataflow entity should more simply. (if you are familiar with power query, you can direct coding formula with Azure AS connector and azure as datasource connection string. Otherwise, you can do these steps in power bi desktop navigation panel and copy backend query table formula to service side)

Regards,

Xiaoxin Sheng

[pawlowski6132]

Thanks for the reply. "Complex Steps" indeed. I don't know which is worse, that or scheduling a model refresh in Azure Analysis Services. For the life of me, I can't understand what it is so complex when, while it is so easy to do in PBI.Com (refresh data model)

[v-shex-msft]

Hi @pawlowski6132,

In fact, create a dataflow entity based on model.json or azure AS connector is equivalent. (In my opinion, model.json mostly used to backup dataflow entity or migrate entity to other workspaces)

BTW, dataflow refresh will get the latest data from the specific data source, if you haven't updated source azure as tabular data model, it still gets the existed older records.

In summary, you still need to refresh on Azure AS tabular to update records. If you feel confused and need some further support about azure AS, I'd like to suggest you post to Azure community.

Azure Community Support 

Regards,

Xiaoxin Sheng