Incremental refresh - publish

[Anonymous]

Community -  I am not clear on this comment below from the documentation.    

Let's say I set up incremental refresh on a dataset.   Publish.   Then I get asked to create a new measure on that dataset, and I re-publish it.   Does this mean the incremenatal refresh is now no longer valid?

Incremental refresh for datasets in Power BI - Power BI | Microsoft Docs

Configuring incremental refresh is done in Power BI Desktop. For most models, only a few tasks are required. However, keep the following in mind:

• When published to the service, you can't publish the same model again from Power BI Desktop. Republishing would remove any existing partitions and data already in the dataset.

[TomMartens]

Hey @Anonymous ,

re-publishing a dataset Power BI Desktop will re-create the initial partitions, meaning an initial load will be necessary. Things can become more and more complex, as on a certain point in time the initial load will no longer succeed due to the amount of data.

It's very complex to use incremental refresh without having a Premium capacity. A Premium capacity provides the feature: XMLA endpoint. The XMLA endpoint can be used by the ALM Toolkit (http://alm-toolkit.com/). The ALM toolkit allows to deploy a measure without overwriting the existing partitions.

Hopefully, this provides some new ideas on how you can tackle your challenge.

Regards,

Tom

[Greg_Deckler · Accepted solution]

@Anonymous Essentially correct. The dataset would be replaced by whatever is in the PBIX file. You can avoid this by using something like ALM Tooklit or Tabular Editor. ALM Toolkit allows you to compare a dataset in the Service (Premium) with a PBIX, identify the model differences and update the model (not the data) with just the changes identified.