SAS Viya Decision Pipeline with Azure DevOps
What if you could combine SAS Intelligent Decisioning with Azure DevOps to create a decision pipeline? A pipeline that imports decisions stored in Git, loads data needed for tests, guides the user to test and publish the decision and finally, deploys the decision. Watch the video demonstration in this post to find out more.
This is the fourth post in the series. The previous posts covered:
- SAS Viya Model Release Pipeline with Azure DevOps
- SAS Viya with Azure DevOps Data Pipeline
- SAS Viya CI/CD Pipelines with Azure DevOps
Requirements
Suppose you have a customer that works with SAS Intelligent Decisioning. The customer requires:
- Automatic processes to import, export and deploy the decision. Decisions are scored by issuing REST API calls.
- An automatic process to load the data needed for tests.
- Easily track the changes from one decision version to another.
- Manual review steps, where the user can test and confirm the test results before deployment.
- Logs and tracking of the above steps.
You could solve the above requirements with SAS Viya and Azure Pipelines.
SAS Viya Decision Pipeline
Watch this short video demonstration:
The decision pipeline has multiple stages. Each of the stage has jobs. Every job takes care of a task and fulfills a part of the requirements. For instance:
- Develop stage: Import a decision from a package file, stored in a Git repository (Azure Repos). Load the test data in Cloud Analytic Services (CAS). Ask the user to test the decision with the test data.
- Publish stage: If the test is successful, publish the decision to a Git publishing destination, such as GitHub. When you publish a decision to Git, you can do two things: first, version the decision files in a human readable format; second, prepare its deployment. After publishing, a job in this stage lists the Git repository folders and the files created.
- Deploy stage: Use the decision files published to Git to deploy to another publishing destination, SAS Micro Analytic Service (MAS). From Git, you can publish to MAS or CAS. Publishing to MAS allows you to score with a REST API.
Select any image to see a larger version.
Mobile users: To view the images, select the "Full" version at the bottom of the page.
Conclusions
The pipeline demonstrated only a subset of what is possible. You can go even further and add other jobs or pipelines before or after. Before, to create publishing destinations. After, to score the decisions deployed and check the results. You can deploy the decision as a container image in Azure, as shown in the SAS Viya Model Release Pipeline with Azure DevOps.
You can combine jobs managing decisions inside SAS Viya with other Azure Pipelines tasks and scripts. You can therefore place decisions developed in SAS Viya in a larger enterprise-wide scope.
Resources
- SAS Viya and Azure Pipelines Self-Hosted Agents
- SAS Viya and Azure Pipelines Microsoft-Hosted Agents
Thank you for your time reading this post. If you liked the post, give it a thumbs up!
Please comment and tell us what you think about this topic. If you wish to get more information, please write me an email.
Find more articles from SAS Global Enablement and Learning here.
