ADF – Deployment from master branch code (JSON files)

ADF – Deployment from master branch code (JSON files)

In the previous episode, I showed how to deploy Azure Data Factory in a way recommended by Microsoft, which is deployment from adf_publish branch from ARM template.
However, there is another way to build CD process for ADF, directly from JSON files which represent all Data Factory objects.
In this video, I will show you the latter method, building the entire process step-by-step, including the fact that we have multiple environments and wherefore isolated services. Therefore, we need to manage all the needed properties, correctly.

Update @ 02/06/2020:

Since now, I recommend this extension to Azure DevOps only, if you plan to deploy ADF from code.
Extension: Deploy Azure Data Factory by SQLPlayer
It is free from objects dependencies issue and does all the work in one task. Check it out.

Update @ 02/05/2020:
Alternatively, you can use Azure PowerShell task in Azure DevOps and use a brand new PowerShell open-source module to publish the whole Azure Data Factory code from your [master] branch or directly from your local machine. The module resolves all pains existed so far in any other solution, including deployment part of objects, deleting objects not existing in the source any longer, stop/start triggers, etc.
More details and full documentation with code is in GitHub:

People are still divided on how to deploy ADF:


The series about ADF on SQLPlayer blog.
Extension to Azure DevOps I used in this episode.

Previous ASF 030: Benni De Jagere interview
Next SSDT: Scripting static data from SQL Server (video)

About author

Kamil Nowinski
Kamil Nowinski 191 posts

Blogger, speaker. Data Platform MVP, MCSE. Senior Data Engineer & data geek. Member of Data Community Poland, co-organizer of SQLDay, Happy husband & father.

View all posts by this author →

You might also like

General 4 Comments

Skirmishes with automated deployment of Azure Data Factory

When it comes to Azure Data Factory deployment, I always have a few words to say. You might be aware that there are two approaches in this terms. I was

Azure Data Factory 8 Comments

Azure Data Factory v2 and its available components in Data Flows

Many of you (including me) wonder about it. Namely: Is it possible to move my ETL process from SSIS to ADF? How can I reflect current SSIS Data Flow business

ASF 020 video invitation

New podcast’s episode is coming (next Friday) and in the meantime, take a look you can watch that episode’s trailer. Do you know which conference was inspired by the SQLBits


  1. ram
    April 23, 20:22 Reply

    Great content and thank you for sharing this.

    Is it possible to just publish only the changed json files with this approach rather than publishing all the json files all the time?

  2. Yash
    June 23, 06:57 Reply

    Great video Kamil.
    Can you please let me know on how we can check on hot-fix part. i.e. We came across an issue in production branch and want to apply the hotfix first in production and then UAT /Dev stage.

    • Kamil Nowinski
      June 23, 10:27 Reply

      Hello Yash and thanks.
      For hot-fixes, you should create a separate branch (e.g. “hot-fix”) based on production (“master”), fix the issue in there and deploy the code into required environments (PreProd). Then those changes should be merged into “master” branch and publish to the Production environment. More details in future videos.

      • Yash
        June 26, 22:54 Reply

        Thanks Kamil… Also, we have multiple developers in our team who need to use same ADF for development purposes. This code will be merged back to master and then go to UAT and so on.

        Each developer is being allocated with his own SQL Server and Storage, If everyone starts creating the separate linked services for development, this number will grow exponentially which might give issues on deploying code to other environments etc.

        Do you have any recommendation on how we can deal with this situation.

        • Kamil Nowinski
          June 27, 09:17 Reply

          Sure. It’s a very good practice to have isolated databases for developers. In that case, each developer should work on its separate branch (ADF) and merge all changes (features, including pipelines, datasets, etc), except e.g. Linked Service(s) whom pointing to its own resources.

  3. Yash
    June 27, 19:19 Reply

    Great .. I was under assumption that by default all components gets merged (including Linked service) and we do not have any control over it. Do I need to use any customized components for excluding linked services from merge.

    • Yash
      June 30, 19:20 Reply

      Hi Kamil, It looks like the developers created datasets will refer their development phase linked service.
      Datasets might give issues on other environments if we wont merge the linked services.
      Do you have any inputs on how we can handle that situation 🙂

  4. successhawk
    September 04, 14:27 Reply

    Hi. Great content. Can you provide a pros and cons list for each approach (Microsoft’s proposed ARM vs. this)? Why would someone choose one over the other? It does appear to me that this is better but I would like to see your experienced opinion.

    • Kamil Nowinski
      September 04, 19:16 Reply

      I will prepare the whole blog post about it. Hopefully, this month yet.

  5. bharani
    October 23, 16:15 Reply

    Hi, I have downloaded and configured your devops plugin. Thanks for your initiative. I am stuck in a place where I have to override service principal key value of a data lake / storage. From Git samples I only see that you are excluding “linkedService,BlobSampleData,-typeProperties.encryptedCredential” encrypted credential. I tried taking the encrypted creds from dev and tried to use the same for test adf but did not work. Is there a way to make this work without manually changing linked service values in test adf.

Leave a Reply