How to implement CI/CD in Azure Data Factory

A CI/CD pipeline is a series of steps that must be performed in order to deliver a new version of software. Continuous integration/continuous delivery (CI/CD) pipelines are a practice focused on improving software delivery using either a DevOps or site reliability engineering (SRE) approach.

These are just an example of common stages you may find. Your pipeline will be unique to the requirements of your organization.

In ADF, CI/CD means moving Data Factory pipelines from one environment to another such Dev –> QA 🡪 Prod.

Create a Resource Groups for different environment with similar resources.

In this example, ADF pipeline used Azure Storge Account & Key vault

Create a new Azure DevOps Project

Crate a new repository

Configure DEV-ADF-U repository to ADFRepo and publish a pipeline.

Make sure all these linked services are using DEV resources.

Once you publish, you can see new branch named called adf_publish branch automatically created in ADFRepo repository.

Inside the adf_publish branch, you can see the ARM template files. One for json file for pipeline and the other josn file for parameter.

Now let’s see how to move DEV data factory to UAT data factory

Under DevOps click on Pipelines 🡪 Releases 🡪 create Pipeline 🡪 click on Empty job

Set stage name as UAT

Rename Pipeline name as appropriate

Click on add an artifact

Configure the artifact values to the release pipeline as sown below.

You can add new stages such as PROD from UAT

Click on UAT Stage Job

Click on Add tasks 🡪 click + on Agent Job 🡪 Search the template using ARM

In this case ARM template deployment is used.

Configure the setting for the UAT environment

First Authorized the Azure Subscription, this will create a service principle for the pipeline, you can see that under App Registration in AAD

The configure the rest of the settings

Select Resource Group as RG-UAT

Select ARMTemplateForFactory.json as Template

Select ARMTemplateParametersForFactory.json as Template parameters

Overwrite the parameter from DEV to UAT

Once you configure all the values, it will be as show below

Click on Save button, this will successfully create a release pipeline for you.

To release to UAT

click on Releases 🡪 ADFReleasePipeline 🡪 click on Create release 🡪 click on create button again

Once you click on it, it will release everything to UAT

In my case, I haven’t purchased or requested the parallelism yet which is required for release, so I got error. You can request it free using a submitting form, it would take 2-3 days to approve.

##[Error 1]

No hosted parallelism has been purchased or granted. To request a free parallelism grant, please fill out the following form

Cheers!
Uma

How to set up code repository for Azure Data Factory using Azure DevOps

First, if you don’t have an Azure DevOps account signup for free. Make sure you select the Default directory.

Create an organization and a project.

Organization - UmaAzure01
Project – azure_adf

Create a new Repository – Imagine a Repository is like a Parent Folder

 

Now go to Azure Data Factory and click on Set up code repository.

Here Collaboration branch means – the Final Code of the application. Usually, the main branch is set as a collaboration branch which is called as Main/Master Branch.

You can create multiple branches which are called a Feature Branch.

Branches will be merged with the master branch through Pull Request. The pull request is for approval for merging with the master.

Create a new working branch or you can use the main. In here ad_dev_branch1

Let’s create a pipeline and test. Create a pipeline under adf_dev_branch1 and saved it.

You can log in to the repository and create the Pull Request to merge

adf_dev_branch1 🡪 main

Approve the Pull Request and click on the complete button.

Once you the is completed, you can see changes in Main Branch in ADF

Cheers!
Uma