Within Power BI, Deployment Pipelines can be used to deploy content to different workspaces. These pipelines are often used to promote reports and/or models from development to, for example, test and production workspaces. Initially, pipelines were only available with a Premium Capacity license, but they are now also available to Premium Per User subscribers.
Method
It works best if you start using the deployment pipelines right from the beginning when developing models and reports. That way, you’ll learn right away how to work effectively and navigate the various stages of development. When creating a pipeline, the phases “Development,” “Test,” and “Production” are suggested by default. These phases can be customized and organized into a sequence ranging from a minimum of 2 to a maximum of 10 phases. For each phase, you can also specify whether it should be made public and thus accessible to users who do not have access to the pipeline. By default, only the last phase of the pipeline will be made public.
When applying changes to the next phase, the system indicates for each object whether there are any differences. You can also choose not to include all objects. Additionally, for the selected objects, you can specify whether the related objects should also be deployed. For semantic models, the differences between the models in the various phases can also be highlighted. Unfortunately, it is not possible to specify for each individual change whether or not it should be included in the next phase. If this is desired, you will need to consider a tool such as ALM Toolkit.
Deployment Rules
When using pipelines, the models are fully synchronized. Of course, there may be times when you want to configure certain settings specifically for each phase. A simple example of this is a database connection. You want the model in development to point to a development environment and the production model to point to production. Deployment rules can be used for these types of scenarios. These deployment rules are only available for semantic models and can be configured on a per-phase basis.
Data Source Rules
The most obvious deployment rules are the data source rules. If desired, you can specify for each phase and each data source where it should connect. These rules are then applied when the model is deployed.
Parameter rules
Parameter rules actually work the same way. They also allow certain parameters to be set differently for each phase. However, these parameters must be specifically designated as parameters within the semantic model. So, for example, this does not work for a standard Shared Expression.
Car Binding
Another way to ensure that reports and models are properly linked is to use autobinding. If the reports and semantic models are in the same workspace, this isn’t necessary, but if they’re in different workspaces, it is. Auto-binding ensures that when reports and models are deployed via different pipelines, the reports in the development phase are automatically linked to the models in the development phase. If you do not want to use autobinding, there are three ways to do so:
- Make sure a report is linked to a model in a different phase. If a report in Development is linked to a model in Production, it will retain that link upon deployment.
- Use parameters. However, this is only possible with models, not with reports.
- Use models that are not linked to a pipeline.
Conclusion
Deployment Pipelines are very useful for managing and developing reports and models in Power BI. The use of deployment rules makes them very easy and flexible to use. It’s just a shame that a model can only be deployed in its entirety. For large models, it’s more convenient to use a different tool so that parts of the model can be deployed separately.
