Fabric Data Factory Pipeline: Incremental load

[waltheed]

How can I create an incremental load in Data Factory Pipeline?

I only see the Table Actions append or overwrite. But I want to be able to add new rows, update existing ones and mark deleted rows. I would expect that you can tell the wizard what the PK columns are and that this logic is created for you.

Is there a way to do this now? Or is this a feature that is coming later on?  

Many thanks!

[cw88]

hi,
is it already known when the function will be generally available? I also need this function.

Thanks!

[ajarora · Accepted solution]

Your table action options imply that you are using pipelines. If so, good news, we have just added a new feature that allows you to "upsert" the new rows into existing tables instead of append/overwrite only.

Its in private preview, please ping me with your workspace id, and we can enable it for you along with instructions. cc @makromer 

[cw88]

@ajarora : do you know when the function will be available for everyone?

[wgjunsay]

Hello, is it possible to have this enabled for our capacity? It would help us a lot as we are planning on moving some of our enterprise data to Fabric from SQL DB. 

[matkvaid]

Hi, is it still possible to try this feature? I have set up incremental refresh with stored procedures etc, but want to have that setup for first step from on prem sql to lakehouse, and would not want to recreate everything again or leave full refreshes for prepared data from lakehouse to warehouse 🙂

[DiegoPLAB]

Hi @ajarora,

do you know when this feature will be general available?

I would like to do upsert with copy data or dataflow instead of stored procedure.

Thanks.

[MonztaInc]

Hi @ajarora 

Is this still in Private Preview or is it in the GA as well? 

[amolt]

Hi ajora, 

When will this action be available outside the private preview?

thanks

[datamind]

Hi @ajarora,

Thanks for providing the information! I have a question about using Data Factory pipelines to upsert data from an on-premise SQL Server incrementally. Is there a method available for doing this?

I've managed to establish a connection through our Gateway to the Server and would like to use the DF Copy activity to query data and upsert it into a Datalakehouse delta table. However, I'm not seeing the created connection in the Copy activity wizard. Is there something I might be missing?

On the other hand, I can connect to it using Dataflows Gen2, apply a filter, and then use the Datalakehouse as the destination for data in append mode. The problem is, I'm not quite sure how to query only new records without loading the entire table initially.

Do you know if there's a recommended method to achieve this? Especially now during the preview period of Fabric, I'd appreciate any guidance you might have.

Thanks in advance for your help!

[ajarora]

Unfortunately currently Gateway based connections are not supported in Pipelines. We are actively working on it. Until then, if you can use dataflows to stage your onprem data to a cloud store (say SQLDb) then use copy to UPSERT it to LH table, that would be a workaround.

[sergs]

Hi

I've pm'ed you our workspacec id. Would you be able to enable the preview feature? We're evaluating different options at the moment and it's one of the requirements to have incremental load.

Thanks!

[miguel]

Hey!

At the moment, we don't have an incremental refresh feature in the same way that Dataflows Gen1 have them in something like Power BI. We are internally tracking this and I've also created an idea so you can vote for it and help us prioritize it:

https://ideas.fabric.microsoft.com/ideas/idea/?ideaid=4814b098-efff-ed11-a81c-6045bdb98602 

There are some alternatives or different approaches to incremental refresh depending on what logic your incremental refresh should follow by leveraging the existing capabilities with pipelines (for the orchestration component) and the output destinations using either an Append method or loading to a staging table and then running any sort of scripts that you'd like to run to make the incremental refresh happen through a notebook.