Practice Exams:

Microsoft AZ-104: Azure Subscription Design

Cloud & Architecture

An Azure subscription is simultaneously a deployment scope, a governance boundary and a unit with billing and service limits. Good subscription design makes those boundaries deliberate without turning every application team into an isolated platform operator. Subscription boundaries should reflect governance, operational ownership, service limits, and cost allocation rather than every application name. A good design lets shared networking and security controls apply consistently without creating needless cross-subscription dependencies.

On this page
  1. Start with the purpose of a subscription boundary
  2. Understand the relationship with tenant and billing account
  3. Use environment and regulatory differences deliberately
  4. Design subscription vending as a controlled product
  5. Plan quotas, capacity and shared-service limits
  6. Control spending without equating budgets with enforcement
  7. Govern lifecycle, transfers and retirement

Start with the purpose of a subscription boundary

An organization might create separate subscriptions for central networking, security services, production applications, test environments and regulated workloads. That separation can support different ownership, budgets, quotas, deployment permissions and governance requirements. Yet a subscription should not be created merely because another cost center appears on an organization chart. A Microsoft AZ-104 subscription boundary is most useful when it separates governance, identity, cost, and operational ownership deliberately.

Ask what must differ between the proposed subscriptions. Do they need separate administrative authority, billing accountability, high-impact change approvals, connectivity, regions or compliance controls? Will one workload exhaust a service quota that another workload depends on? If the answer is only “we want to see costs per team,” resource tags, budgets and reporting may sometimes be enough inside a shared subscription.

The subscription boundary cannot replace application tenancy controls. Two applications in separate subscriptions may still use the same Entra tenant, central firewall, shared DNS zone or logging workspace. Document those shared dependencies so an application owner does not assume that billing separation implies complete security or outage isolation.

Understand the relationship with tenant and billing account

A subscription has a relationship with a Microsoft Entra tenant for identity and Azure RBAC, and it is also associated with a billing arrangement. These are distinct organizational constructs. Billing roles, tenant-level directory administration and subscription-scoped Azure resource roles are not automatically interchangeable. Verify who can view invoices versus who can deploy or delete resources.

A subscription owner may manage Azure resources while the billing account holder controls purchasing or billing scope. Changing the subscription’s directory or offer can have significant operational effects, including identity mappings and access, and should not be treated as a routine cost-center edit. Record tenant association, subscription ID, offer/billing context and billing owner in a canonical inventory.

For access design, place the subscription under the appropriate governance hierarchy and assign workload team roles at the smallest practical scope. Azure management groups organize shared policy and role inheritance above individual subscriptions. Their hierarchy should tell a platform team which baseline controls a new subscription will receive before anyone deploys its first virtual machine.

Use environment and regulatory differences deliberately

Consider two departments that both deploy virtual machines. The first runs temporary nonproduction analytics and tolerates a weekly rebuild. The second operates a regulated financial service with production change windows and longer retention. A single shared subscription could technically host both, but its governance rules would become a tangle of exceptions and overbroad role assignments. Separate production subscriptions within purpose-built management-group branches may express these differences more clearly. Conversely, splitting every experimental notebook into another subscription adds administration without improving the policy boundary. The decision turns on controls and risk, not on counting resource owners.

Production and development commonly justify different change controls, budgets, identity permissions and connectivity routes. Separating them at subscription level can reduce accidental impact and simplify governance, but only if related shared services and network paths are also addressed. A development subscription attached directly to a production hub without controls can still affect central services.

Regulated workloads may require different permitted regions, data access policies, retention settings and incident evidence. If those requirements materially differ from standard workloads, a dedicated subscription or management-group branch can make policy enforcement clearer. Avoid assuming a separate subscription provides compliant data residency by itself; check where backups, diagnostic logs, managed keys and dependent SaaS services store data.

Sandbox environments need explicit expiration, quota and spending rules. A separate sandbox subscription is easier to isolate but can create subscription sprawl if no one retires it. Decide how experiments are provisioned, who approves temporary exceptions, when idle resources are cleaned up and whether developers may connect the sandbox to sensitive network or identity systems.

Design subscription vending as a controlled product

Manual subscription creation often leaves each new environment with slightly different policies, owners and monitoring. Subscription vending provides a standard workflow for requests, approval, creation, placement and baseline configuration. Microsoft’s subscription vending guidance describes a repeatable pattern for gathering requirements and automating the platform steps.

A request should include workload name, environment, business owner, data classification, budget responsibility, required regions, connectivity needs and expected lifecycle. An approval should validate the need for a new boundary. Automation can then create or enroll the subscription, attach it to the intended management group, assign limited administrative roles, set required controls and register the resulting subscription ID in the estate inventory.

A complete vending workflow also proves the controls took effect. Before handing a subscription to an application team, verify policy assignments, log destinations, required security protections, network connectivity, contact details, tag standards and cost alerts. Simply receiving a new subscription ID is the beginning of the service, not the end of onboarding.

Plan quotas, capacity and shared-service limits

Suppose an application needs 80 additional vCPUs in the West Europe region for a weekend migration and a recovery environment in another region. The platform team should verify the exact family quota, deployed and available capacity constraints, subscription owner and any reserved public IP or network appliance limits before approving the schedule. A quota approved in the primary region does not prove the secondary region can support the same deployment. This is particularly important for a disaster-recovery subscription that sits idle during normal months: lack of recent deployments makes its quotas and dependencies easy to overlook.

Azure services expose different quota and limit models by subscription, region and SKU. Some limits can be increased through approved support processes; others are structural. Review Azure service limits for the particular resources the workload will use. Do not assume a new subscription removes a regional capacity shortage or that every service permits unlimited quota increase.

An application running many VMs may need sufficient vCPU quota in the actual target region, while networking infrastructure can depend on public IP, gateway or address-space constraints. Reserve capacity and perform quota checks before a large migration or disaster recovery drill. In an emergency, a spare subscription with insufficient quota may be no more useful than the damaged primary environment.

Shared platform costs can also be hidden from application subscriptions. Central network egress, DNS, logging and security services are often billed elsewhere. Define how these costs will be allocated or reported so a team does not assume the workload’s own subscription invoice represents its complete cloud consumption.

Control spending without equating budgets with enforcement

Budgets and cost alerts help an owner track spending, forecast variance and respond early. They do not always stop deployments or resource consumption when a threshold is crossed. A responsible design pairs budget reporting with pricing awareness, policy constraints where applicable, resource lifecycle automation and owner accountability.

Define costs at the subscription, resource group or supported billing scopes appropriate to the organization. Adopt consistent tags for workload, environment, accountable owner and financial classification. Some cost categories remain untagged or indirectly attributable, so reports should expose that uncertainty instead of forcing a misleading percentage of “allocated” spend.

Review anomalies with operational context. A spike may reflect expected load, an incident, a rollout, unexpected data transfer or an orphaned resource. The investigation process in Azure cost governance should connect cost evidence to a particular workload change before concluding that a team overspent.

Govern lifecycle, transfers and retirement

An effective retirement checklist follows dependency ownership instead of simply looking for running compute. Remove or transfer DNS records, policy exemptions, managed identities and monitoring integrations; verify that certificates and secrets no longer point to retiring endpoints. Check that backups and retained data have an approved legal and security disposition, and that shared network routes and private endpoints have new owners where necessary. A subscription can look empty in a basic resource list while its identity assignments, cost commitments or dependent external integrations still matter. Closing the subscription is the final recorded step, after evidence is reviewed.

A subscription may move to a different workload owner, management group or billing arrangement over time. Such transfers can change inherited policy, access or financial controls even if the contained resources stay untouched. Use change management to compare before-and-after effective policy, roles, quotas, connected services and billing recipients.

Retirement needs more than deleting a subscription or stopping payment. Identify active workloads, backups, retained records, encryption keys, private DNS dependencies, shared service links and legal hold requirements. Determine what must be exported or kept and what can be removed under approved retention policy. Disable broad access only after the recovery and records strategy is confirmed.

A subscription has a useful purpose when a named team owns it, billing and quotas match that purpose, inherited controls are visible, and its lifecycle can be managed without damaging unrelated applications. Test onboarding and offboarding as carefully as steady-state access. Creating more subscriptions is not inherently better governance if services continue to share undocumented privileges or network dependencies.

Continue learning

Related guides

Azure Management Groups at ScaleApply inherited governance across Azure management groups and subscriptions.Cost Governance for Azure SubscriptionsUse budgets, cost allocation, and owner accountability for spend.Subscriptions, Azure Policy, and Resource LocksSee how subscriptions, Azure Policy, and locks constrain deployments.

Related Posts

• Legacy MCSA Cloud Platform: Azure Architecture Skills and Current Paths

• Microsoft AZ-104: Is Azure Administration the Right Career?

• Azure Mastery Beyond the Basics: Governance, Green Cloud, and Strategic Growth

• Cloud Strategy Unlocked: How AZ-305 Shapes Real Azure Architects

• Microsoft AZ-104: Managed Identities in Azure

• Microsoft AZ-104: Entra Roles and Access

• Microsoft SC-300: Hybrid Identity: Sync, Federation, and Cloud Auth

• Microsoft AZ-900: Cloud Economics: CapEx, OpEx, and the Cost of Commitment

• Microsoft AZ-104: Zones and Failure Domains

• Microsoft AZ-104: FSLogix Profiles for Azure Virtual Desktop