Read a sample: 1. The Deletion Decision
The retirement request looks simple: “The old reporting environment is no longer needed. Please delete its resource group.”
An administrator finds a group named rg-reporting-dev and sees a handful of resources. The name suggests development, the request came from a familiar team, and the portal offers a delete action. It would be easy to treat the task as routine housekeeping.
But the name does not establish ownership. A development environment can still contain a copy of data subject to a retention requirement. A network resource inside the group may serve something outside it. A backup may exist without having been tested. The person asking for deletion may know the application is retired but not know that another team uses one of its components.
None of these possibilities means the group must be kept. They mean that deletion is a decision before it is an operation.
This chapter introduces the boundary an Azure resource group provides, the broader effects that boundary does not capture, and the decision gates used throughout this book. The goal is not to make every small cleanup a lengthy investigation. It is to make the depth of review proportional to what could be lost or disrupted—and to leave a clear record of why the action was authorized.
An Azure resource group is a management container for related Azure resources. Resources such as virtual machines, storage accounts, virtual networks, and other service instances can be organized into groups so that administrators can manage their lifecycle, access, and governance with a useful scope.
A resource group belongs to a subscription. An individual Azure resource is associated with one resource group at a time, although the services that resource uses or supports may be located in other groups or subscriptions. The group is therefore an administrative boundary, not necessarily an application boundary, a network boundary, or a complete record of business ownership.
That distinction matters because organizations arrange resources in different ways. One team might place nearly everything for an application in a single resource group. Another might separate resources by environment, owner, lifecycle, or technical function. Shared networking and monitoring may sit outside an application’s group. Conversely, an application’s group may contain a resource that has acquired consumers the original designers never anticipated.
Resource groups also provide a scope at which Azure governance controls can apply. Access can be granted at the group level or inherited from a broader scope. Management locks can protect a group or its resources against certain operations. Policies and organizational procedures may also shape what administrators are permitted to do. These controls are reasons to investigate, not obstacles to remove automatically.
A useful mental model is a labeled container on a crowded workbench. The label helps you find the contents, but it does not tell you which other tools are connected to them, which items must be preserved, or who is authorized to discard them.
Deleting a resource group initiates deletion of the group and the resources within it, subject to Azure’s processing and any applicable controls or service-specific behavior. It is not merely the removal of a label. A deletion request can start a sequence of operations against many resources, and those operations may take time or encounter errors.
The effects are not always neatly confined to the visible list in the group. A resource outside the group may remain in Azure but lose a dependency it needs. A service-specific retention or recovery feature may preserve some data after deletion, while other data or configuration may be lost. External systems may continue to refer to endpoints, identities, or resource identifiers that no longer work.
The distinction is between the deletion boundary and the impact boundary:
- The deletion boundary is the target resource group and the resources Azure processes as part of deleting it.
- The impact boundary includes the people, workloads, data, integrations, and operational processes affected by removing those resources, whether or not they are in the group.
The impact boundary must be investigated; it cannot be inferred from the group name alone. Finding an outside dependency does not necessarily prohibit deletion. It may mean that a consumer must be migrated, notified, disconnected, or explicitly accepted as part of the retirement.
Nor should an administrator assume that a group is empty merely because the most obvious application has stopped. Supporting resources can remain after a workload is retired. Some may have independent value or obligations; others may be abandoned and suitable for removal. An inventory starts that conversation, but it does not make the disposition decision on its own.
Consider a fictional organization, Northstar Outfitters. Its analytics team submits a request to retire an old reporting environment in Azure. The request identifies rg-reporting-dev, says the replacement reporting platform is live, and asks cloud operations to delete the old environment to reduce clutter and ongoing costs.
Maya, the administrator assigned to the request, resists the urge to begin with the delete button. She first asks what the requester means by “old environment.” Is the target only the resource group named in the ticket, or does the team expect related resources elsewhere to disappear too? Is this a development environment in practice, or only in its name? Has the business owner agreed that the data and services can be retired?
The initial review reveals three issues that need owners before the team can proceed.
First, there are similarly named groups in different subscriptions.…