Design Azure Policy as Code workflows

As you progress on your journey with Cloud Governance, you'll want to shift from manually managing each policy assignment in the Azure portal or through the various SDKs to something more manageable and repeatable at an enterprise scale. Two of the predominant approaches to managing systems at scale in the cloud are:

• Infrastructure as Code: The practice of treating the content that defines your environments, everything from Azure Resource Manager templates (ARM templates) to Azure Policy definitions to Azure Blueprints, as source code.
• DevOps: The union of people, process, and products to enable continuous delivery of value to our end users.

Azure Policy as Code is the combination of these ideas. Essentially, keep your policy definitions in source control and whenever a change is made, test and validate that change. However, that shouldn't be the extent of policies involvement with Infrastructure as Code or DevOps.

The validation step should also be a component of other continuous integration or continuous deployment (CI/CD) workflows, like deploying an application environment or virtual infrastructure. By making Azure Policy validation an early component of the build and deployment process, the application and operations teams discover if their changes are behaving as expected long before it's too late and they're attempting to deploy in production.

Definitions and foundational information

Before getting into the details of Azure Policy as Code workflow, it's important to understand some fundamental concepts, like how to author policy definitions and initiative definitions, and how to leverage exemptions on assignments of those definitions:

• Policy definition
• Initiative definition
• Policy exemption

The file names correspond with certain portions of policy or initiative definitions and other policy resources:

Examples of these file formats are available in the Azure Policy GitHub Repo

Workflow overview

The recommended general workflow of Azure Policy as Code looks like this diagram:

Source control

Existing policy and initiative definitions can be exported different ways such as through PowerShell, CLI, or Azure Resource Graph (ARG) queries. The source control management environment of choice to store these definitions can be one of many options, including a GitHub or Azure DevOps.

Create and update policy definitions

The policy definitions are created using JSON, and stored in source control. Each policy has its own set of files, such as the parameters, rules, and environment parameters that should be stored in the same folder. The following structure is a recommended way of keeping your policy definitions in source control.

.
|
|- policies/  ________________________ # Root folder for policy resources
|  |- policy1/  ______________________ # Subfolder for a policy
|     |- policy.json _________________ # Policy definition
|     |- policy.parameters.json ______ # Policy definition of parameters
|     |- policy.rules.json ___________ # Policy rule
|     |- assign.<name1>.json _________ # Assignment 1 for this policy definition
|     |- assign.<name2>.json _________ # Assignment 2 for this policy definition
|     |- exemptions.<name1>/__________ # Subfolder for exemptions on assignment 1
        | - exemptionName.json________ # Exemption for this particular assignment
      |- exemptions.<name2>/__________ # Subfolder for exemptions on assignment 2
        | - exemptionName.json________ # Exemption for this particular assignment
|
|  |- policy2/  ______________________ # Subfolder for a policy
|     |- policy.json _________________ # Policy definition
|     |- policy.parameters.json ______ # Policy definition of parameters
|     |- policy.rules.json ___________ # Policy rule
|     |- assign.<name1>.json _________ # Assignment 1 for this policy definition
|     |- exemptions.<name1>/__________ # Subfolder for exemptions on assignment 1
        | - exemptionName.json________ # Exemption for this particular assignment
|

When a new policy is added or an existing one is updated, the workflow should automatically update the policy definition in Azure. Testing of the new or updated policy definition comes in a later step.

Create and update initiative definitions

Initiative definitions are also created using JSON files that should be stored in the same folder as policy definitions. The initiative definition requires the policy definition to already exist, so it can't be created or updated until the source for the policy has been updated in source control and then updated in Azure. The following structure is a recommended way of keeping your initiative definitions in source control:

.
|
|- initiatives/ ______________________ # Root folder for initiatives
|  |- init1/ _________________________ # Subfolder for an initiative
|     |- policyset.json ______________ # Initiative definition
|     |- policyset.definitions.json __ # Initiative list of policies
|     |- policyset.parameters.json ___ # Initiative definition of parameters
|     |- assign.<name1>.json _________ # Assignment 1 for this policy initiative
|     |- assign.<name2>.json _________ # Assignment 2 for this policy initiative
|     |- exemptions.<name1>/__________ # Subfolder for exemptions on assignment 1
        | - exemptionName.json________ # Exemption for this particular assignment
      |- exemptions.<name2>/__________ # Subfolder for exemptions on assignment 2
        | - exemptionName.json________ # Exemption for this particular assignment
|
|  |- init2/ _________________________ # Subfolder for an initiative
|     |- policyset.json ______________ # Initiative definition
|     |- policyset.definitions.json __ # Initiative list of policies
|     |- policyset.parameters.json ___ # Initiative definition of parameters
|     |- assign.<name1>.json _________ # Assignment 1 for this policy initiative
|     |- exemptions.<name1>/__________ # Subfolder for exemptions on assignment 1
        | - exemptionName.json________ # Exemption for this particular assignment
|

Like with policy definitions, the workflow should automatically update the initiative definition in Azure when an existing initiative is added or updated. Testing of the new or updated initiative definition comes in a later step.

Test and validate the updated definition

Once automation has taken your newly created or updated policy or initiative definitions and made the update to the object in Azure, it's time to test the changes that were made. Either the policy or the initiative(s) it's part of should then be assigned to resources in the environment farthest from production. This environment is typically Dev.

The assignment should use enforcementMode of disabled so that resource creation and updates aren't blocked, but that existing resources are still audited for compliance to the updated policy definition. Even with enforcementMode, it's recommended that the assignment scope is either a resource group or a subscription that is specifically for validating policies.

After the assignment is deployed, use the Azure Policy SDK, the Azure Pipelines Security and Compliance Assessment task, or Azure Resource Graph (ARG) queries (see samples) to get compliance data for the new assignment. The environment used to test the policies and assignments should have resources with varying compliance states. Like a good unit test for code, you want to test that resources are evaluated as expected with no false-positives or false-negatives. If you test and validate only for what you expect, there may be unexpected and unidentified impact from the policy. For more information, see Evaluate the impact of a new Azure Policy definition.

Enable remediation tasks

If validation of the assignment meets expectations, the next step is to validate remediation. Policies that use either deployIfNotExists or modify can have an associated remediation task triggered to correct resources from a non-compliant state and bring them into compliance.

The first step to remediating resources is to grant the policy assignment the role assignment defined in the policy definition. This role assignment gives the policy assignment managed identity enough rights to make the needed changes to make the resource compliant.

Once the policy assignment has appropriate rights, use the Policy SDK to trigger a remediation task against a set of resources that are known to be non-compliant. Three tests should be completed against these remediated tasks before proceeding:

• Validate that the remediation task completed successfully
• Run policy evaluation to see that policy compliance results are updated as expected
• Run an environment unit test against the resources directly to validate their properties have changed

Testing both the updated policy evaluation results and the environment directly provide confirmation that the remediation tasks changed what was expected and that the policy definition saw the compliance change as expected.

Update to enforced assignments

After all validation gates have completed, update the assignment to use enforcementMode of enabled. It's recommended to make this change initially in the same environment far from production. Validate that the desired effects are applied during resource creation and resource update. Once that environment is validated as working as expected, the change should then be scoped to include the next environment, and so on, until the policy is deployed to production resources.

Process integrated evaluations

The general workflow for Azure Policy as Code is for developing and deploying policies and initiatives to an environment at scale. However, policy evaluation should be part of the deployment process for any workflow that deploys or creates resources in Azure, such as deploying applications or running ARM templates to create infrastructure.

In these cases, after the application or infrastructure deployment is done to a test subscription or resource group, policy evaluation should be done for that scope checking validation of all existing policies and initiatives. While they may be configured as enforcementMode disabled in such an environment, it's useful to know early if an application or infrastructure deployment is in violation of policy definitions early. This policy evaluation should therefore be a step in those workflows, and fail deployments that create non-compliant resources.

Review

This article covers the general workflow for Azure Policy as Code and also where policy evaluation should be part of other deployment workflows. This workflow can be used in any environment that supports scripted steps and automation based on triggers.

Next steps

• Learn about the policy definition structure.
• Learn about the policy assignment structure.
• Understand how to programmatically create policies.
• Learn how to get compliance data.
• Learn how to remediate non-compliant resources.
• Under how to follow policy safe deployment practices