Practice Exams:

Microsoft AZ-104: Resource Groups in Practice

Cloud & Architecture

Azure resource groups are more than folders in a portal. Their boundaries shape what teams can change, what can be retired safely, where policy and permissions apply, and how deployment history can be explained after something breaks.

On this page
  1. Choose boundaries by the work that changes together
  2. Keep location and networking separate from group membership
  3. Design access around operators rather than organizational labels
  4. Policy and locks need a useful scope, not the widest scope
  5. Treat deployment history as an operational record
  6. Use tags to answer business questions the group cannot
  7. Give shared platform components their own accountability
  8. Moving a resource is not the same as renaming a folder
  9. Test group boundaries with an approved change and access request

Imagine retiring a customer application while its shared hub network, DNS, and monitoring remain essential to other services. If everything sits in one resource group, a routine cleanup request can cross ownership and deletion boundaries. An Azure resource group should reflect which resources change and retire together, not just where someone prefers to see them in the portal. In the Microsoft AZ-104 exam context, test that design by tracing one deployment, one role assignment, and one decommission request through the proposed groups.

Choose boundaries by the work that changes together

Picture a service with a public web application, a database, a private network connection and a central monitoring destination. Placing everything into one resource group looks tidy on a diagram. But the items may have different owners, change windows and retirement dates.

When the web application is replaced, the shared network and monitoring systems may need to continue serving other applications. A single container can hide that difference until someone attempts a cleanup.

Microsoft’s Azure Resource Manager recommends grouping resources that share a lifecycle: deploy, update and delete them together. Treat that as the default question, not a rigid rule that every application needs exactly one group. A working boundary expresses a real management relationship. If the application, its supporting compute and its dedicated storage are deployed and retired as a unit, they may belong together even when several resource types are involved.

Conversely, a centrally operated virtual network, private DNS service or security appliance usually has a longer lifetime than one consuming workload. Placing it inside the application’s group would let the application’s lifecycle become entangled with the shared platform. Designing by change ownership often leads to separate groups for a workload, shared connectivity and platform monitoring, with connections between them kept explicit.

The design test is simple: suppose an approved change removes this resource group tomorrow. Would that action retire a coherent service, or would it unexpectedly damage other services that still have owners? If the answer needs a long list of exceptions, the management boundary is probably doing too many jobs.

Keep location and networking separate from group membership

A resource group is a management container. It is not a subnet, a VNet, a firewall policy or an availability zone. Two VMs placed in one group do not acquire an automatic network path just because their Azure resource records sit together. Equally, services in different resource groups can communicate when routing, name resolution and access controls permit it. Group membership says little about packet flow.

Consider an application VM and its storage account. Putting them together does not create private storage access, authorize the VM’s managed identity, or ensure that a storage firewall allows the traffic. A real Azure network troubleshooting follows the endpoint, DNS answer, routes and filtering; a data-access explanation identifies the principal and the permission that allows the operation. Resource-group structure helps administrators locate and govern those components but does not implement the path by itself.

Region labels can produce another mistaken inference. The resource group’s location determines where its management metadata is stored. Resources within it can reside in other supported regions.

Choosing a group location does not move an application, confer data residency on every member, or change the region of a VM. Workload locality should be verified on the individual resource and against the service’s regional constraints.

This distinction matters during reviews. A proposed group named prod-west may contain east-region services and a private network operated elsewhere. The name can be a useful hint, but it is not proof. Check each deployed resource’s actual region, owner and network topology before assessing compliance or resilience.

Design access around operators rather than organizational labels

Resource groups can be Azure RBAC assignment scopes. That makes grouping a practical security decision: granting a team a role at a group can affect many resources underneath it. The correct question is not merely whether the team’s name appears in the group’s name. It is whether every covered resource is legitimately within that team’s operational responsibility and whether that specific role grants more capability than the work requires.

For example, a team maintaining a dedicated application may need to restart its VM and deploy an App Service change. It should not inherit broad permissions over a shared hub network only because someone placed both workloads in one group.

The practical problem of Azure RBAC inheritance also matters: parent-scope assignments can apply from a subscription or management group, so inspecting only a group’s direct assignments is not a complete access review. Microsoft’s Azure RBAC scopes explain the hierarchy and why narrower assignment scopes limit unintended reach.

A resource-group boundary is not, on its own, an application data-permission model. Being able to manage an Azure Storage resource does not imply the same rights to read blobs through an Azure data-plane role. Some management permissions can expose account keys, which makes broad roles particularly sensitive. Design owners and role assignments with the actual action in mind, then validate effective access rather than assuming a group’s tidy layout provides least privilege.

When exceptions proliferate—one member needs a different owner, a second member has restricted operators, a third is shared with another team—the group may be combining responsibilities that should be separated. A small number of coherent groups can produce simpler, more reviewable access than one large group followed by dozens of carefully targeted exceptions.

Policy and locks need a useful scope, not the widest scope

Azure Policy and resource locks are separate controls. A policy assignment evaluates resources against a definition at the relevant scope and effect; a lock protects specified management operations. Neither is equivalent to an Azure RBAC permission grant. Because assignments and locks can apply through hierarchy, the resource-group boundary helps determine which resources inherit a control and which operational changes may unexpectedly fail.

Suppose a production data service requires a delete lock while test VMs in the same group are regularly rebuilt. A delete lock at that group’s scope would complicate legitimate test cleanup. The issue is not that locks are bad: it is that the members do not share the same protection requirement.

The same problem arises when a restrictive policy suitable for one workload creates needless exceptions for another. Review the exception pattern before weakening a valid control for every resource.

Microsoft documents CanNotDelete and ReadOnly locks with importantly different behavior. A delete lock still allows authorized updates while blocking deletion; a read-only lock can block updates and some operational tasks whose underlying control-plane calls are writes. Locks apply to the management plane, not every possible way to modify data. A team should test the actual lifecycle action it intends to protect and should record who can modify the lock configuration.

A well-scoped group makes governance evidence understandable. Its policies, locks, compliance results and role assignments can be reviewed against one declared operational purpose. If that purpose cannot be stated without contradictions, moving individual exceptions around is unlikely to solve the underlying design problem.

Treat deployment history as an operational record

Resource groups are common targets for ARM template and Bicep deployments. If the group corresponds to one owned workload, the deployment history is easier to interpret: a change can be tied to a pipeline, an approver and the resources expected to change together. That is useful during an incident because the investigator can narrow the question from “what changed somewhere in the subscription?” to “which deployment touched the affected workload?”

There is no Azure rule that one template must map to exactly one group. Shared platforms, cross-group references and subscription-scope deployments are legitimate. The operational requirement is traceable ownership. If unrelated pipelines constantly update the same group, record which pipeline owns each resource and be explicit about boundaries. Otherwise a deployment that looks safe from one team’s perspective can unexpectedly interfere with another team’s configuration.

A Bicep file or ARM template describes desired resource configuration; the group identifies one management and deployment scope. Those are complementary but different concerns. A clean group name does not make destructive deployments safe. Microsoft’s deployment modes documentation explains how deployment behavior affects resource removal. Confirm which deletion and deployment options your current tooling supports rather than assuming that a resource group’s tidy layout makes destructive changes safe. Use a what-if review where supported and evaluate destructive scope before applying a change.

For exam preparation, a disposable exercise is to create a group, deploy a simple resource, inspect the resulting deployment record and run a what-if comparison after a harmless parameter change. Explain exactly which property is expected to change and which resource should remain untouched. This teaches the purpose of the management boundary without claiming that the group itself supplies version control or a rollback mechanism.

Use tags to answer business questions the group cannot

Departments, cost centers, environments and data classifications do not always align with deployment lifecycles. Trying to encode all of them in a resource-group tree makes the hierarchy hard to operate. A group has a single position in a subscription, while a resource may need several meaningful business attributes. Names can support recognition; tags can describe the attributes without changing the ownership boundary.

A workload in rg-customer-api-prod might have tags such as Owner, Environment and CostCenter. Those tags support inventory, reporting or policy evaluation where supported, but they do not authorize access or guarantee cost attribution on every usage record. Microsoft’s Azure tags say that resources do not automatically inherit a parent resource group’s tags. Inheritance requires an explicitly configured mechanism such as appropriate Azure Policy or a supported Cost Management setting for usage records.

An important detail is that tagging for governance and inheriting tags for cost reports are not identical operations. A Cost Management setting can attach inherited values to usage data without writing those tags onto the underlying resources. Do not assume that observing a cost allocation value proves the corresponding resource has the tag. Similarly, a tag called Confidential cannot itself restrict which identity reads the resource.

If a team proposes creating a new group solely because the chargeback cost center changed, consider updating approved metadata instead. Reorganize the management container when the ownership or lifecycle boundary has actually changed. That separates a bookkeeping adjustment from a potentially disruptive resource move.

Give shared platform components their own accountability

The most revealing example is a hub-and-spoke environment. A central team owns hub networking, shared DNS and routing components, while individual application teams own their workloads in spokes. Placing the hub inside one application’s resource group creates a misleading deletion and access boundary. The application team may be able to modify infrastructure that other workloads depend on, and removing the application could endanger shared services.

Separate groups can express platform ownership without preventing legitimate cross-group dependencies. In the design record, list the shared dependency, its operational owner, the applications that consume it and the interface they rely on. Network peering, DNS links, private endpoints and monitoring destinations can cross group boundaries by design. The management separation should make that relationship visible, not conceal the dependency.

Not everything shared requires a new resource group. A small environment may still have a coherent, centrally operated platform group containing multiple services with the same owner and lifecycle. Conversely, production and disposable environments that share a team may warrant different groups because their protective controls and expected deletion behavior differ. The decision follows administration, not a rule about the number of services per container.

Moving a resource is not the same as renaming a folder

Azure supports moving many resource types to another resource group, but not every service or configuration supports every move. The operation must be validated for the actual resource type and dependencies. A move within or across groups is a management change, not a region migration; moving a resource between groups does not itself move its physical deployment to another Azure region.

The resource ID contains the group path, so a successful move changes that ID. References in automation, dashboards, alert scopes or templates may then need updates. Microsoft’s resource moves also warns that source and destination groups are locked against resource create, update and delete operations while the move runs, although existing resources remain operational. Changes to role assignments can require remediation after the move; do not presume all inherited and directly scoped permissions follow unchanged.

Before a move, collect the existing resource IDs and dependency list. Check supported move types, quotas, required permissions, locks, provider registration if crossing subscriptions, and applicable tenant restrictions. Record the intended destination owner and re-evaluate group-level RBAC, policy, monitoring and automation references. Validate the plan through the supported Azure move workflow before a production change window, and document a recovery path if dependent systems fail afterward.

The purpose of this checklist is to prevent an organizational tidy-up from becoming an access or availability incident. If the group was created around a clear lifecycle from the start, such moves should be exceptional. When a move is necessary, measure its effects beyond the portal’s success message.

Test group boundaries with an approved change and access request

Imagine a fictional service consisting of an application VM, its dedicated database, central DNS and a shared logging workspace. The application team requests permission to redeploy only its application, while the platform team wants the shared services protected from deletion. Write the proposed resource groups, owners and RBAC scope assignments before opening the portal. If the application team can affect central DNS through a group-wide assignment, the boundary has failed the request.

Also test whether a support engineer can receive temporary diagnostic access without acquiring permissions to change production storage data; verify the effective role and scope, not the resource-group name.

Next, test a disposable version of the design. Confirm that the application team’s role can perform the intended operation on its own resources but not on shared infrastructure. Inspect the relevant policies and locks, apply a harmless tagged deployment and check the resulting deployment history. Then consider what would happen if the application group were deleted after formal retirement, without actually performing an unapproved production deletion.

Before approving a resource-group boundary, ask whether its contents share an owner, change schedule, access policy and retirement plan. A small application may need separate resource groups for network foundations and disposable workload components, even when both support the same service. Validate the design by tracing one deployment, checking inherited policy and role assignments, and identifying which resources would survive an application shutdown. A safe decommissioning test should reveal dependencies, locks, backup retention and shared services before anyone issues a deletion command. That evidence tells an administrator whether the boundary works; a naming convention alone does not.

Continue learning

Related guides

Subscriptions, Azure Policy, and Resource LocksSee how subscriptions, Azure Policy, and locks constrain deployments.Azure Resource Locks and Operational SafetyUse Azure resource locks without blocking authorized recovery.Azure Management Groups at ScaleApply inherited governance across Azure management groups and subscriptions.

Related Posts

• Microsoft AZ-104 Exam Guide: Skills, Labs, and Preparation

• Unlocking Cloud Potential: A Beginner’s Journey into Microsoft Azure Certifications

• Unveiling Microsoft Azure: The Future of Intelligent Cloud Solutions

• Microsoft AZ-104 vs. AZ-900: Which Should You Take?

• Microsoft AZ-104 vs. AZ-103: What Changed?

• Cloud Guardian: Unlocking Your Potential with Microsoft’s AZ-500

• Microsoft AZ-104: Troubleshoot VNets, Routes, and DNS

• Microsoft AZ-104: Managed Identities in Azure

• Microsoft AZ-104: Entra Roles and Access

• Microsoft AZ-104: Azure Landing Zones