Azure Regions, Zones & the Resource Hierarchy

AZ-900 AZ-900 · Describe Azure architecture and services

By Abu Anns · Last updated 20 November 2026Exam version: AZ-900 (skills measured as of July 20, 2026)

How Azure is organised physically

Azure is a global network of datacenters, and the exam expects you to describe how that physical footprint is organised: datacenters, regions, region pairs, availability zones, and sovereign regions. These are the building blocks everything else sits on — compute, storage, and networking all live somewhere, and "somewhere" in Azure means a region, and often a zone inside it.

A region is a geographical area containing one or more datacenters, connected by a low-latency network. Azure has more than 60 regions worldwide — more than any other cloud provider — so customers can usually place resources near their users. When you create a resource, you pick a region, and that choice affects latency (how fast users reach it), cost (prices vary by region), and which services are available (not every service runs in every region). Regions are the unit of data residency: a resource's data stays in its region unless you deliberately configure replication elsewhere.

A datacenter is the physical building: servers, storage, networking, power, and cooling, run by Microsoft. You never interact with a datacenter directly — you interact with regions and zones — but the exam wants you to know datacenters are the physical layer underneath regions.

Region pairs and sovereign regions

Every Azure region is paired with another region — its region pair — at least 300 miles away, inside the same geography where possible (for example, UK South pairs with UK West). Region pairs matter for two reasons. First, platform updates: Microsoft rolls updates out to one region of a pair at a time, so a bad update cannot take down both at once. Second, disaster recovery: services that replicate across regions (such as geo-redundant storage) replicate to the paired region by default, and data residency is preserved because the pair stays in the same geography. If a question describes a service replicating to a second region automatically, the paired region is the destination.

Sovereign regions are isolated instances of Azure for governments with specific requirements — for example, Azure Government in the United States, and Azure China (operated by 21Vianet). They are physically and logically separated from the public Azure cloud and are available only to qualifying government customers. The exam only requires you to know they exist and why: data sovereignty and government compliance.

Availability zones: surviving a datacenter failure

An availability zone is a physically separate datacenter (or group of datacenters) within a region, with its own independent power, cooling, and networking. Regions that support zones have at least three of them. The point of zones is to survive a datacenter-level failure: if you run two virtual machines in different zones and one zone goes down, the other keeps serving. This is the mechanism behind the strongest VM availability SLA in Azure.

Zones are often confused with availability sets, and the exam tests the difference. An availability set protects against failures inside a single datacenter: it spreads virtual machines across fault domains (groups of VMs sharing a power source and network switch, so a hardware failure hits only one domain) and update domains (groups rebooted together during maintenance, so only some VMs restart at once). Zones protect against the whole datacenter failing; sets protect against rack-level failures and maintenance reboots. The SLAs tell the story:

ConfigurationVM availability SLA
Single VM with premium SSD (no redundancy)99.9%
Two or more VMs in one availability set99.95%
Two or more VMs across availability zones99.99%

Remember the order: zones beat sets, sets beat a single VM. A question asking for the highest availability for virtual machines is answered with availability zones.

The resource hierarchy: how Azure is organised logically

On top of the physical footprint sits a logical hierarchy that controls how resources are organised, secured, and billed. From the top:

Management groups sit at the top. They are containers for subscriptions, used to apply governance — policies, access control, compliance — across many subscriptions at once. A management group can contain other management groups, nesting up to six levels below the root, so a large organisation can mirror its structure (for example, a group per division, with groups per department beneath). Whatever you apply at a management group flows down to everything beneath it.

Subscriptions are the unit of billing and access. Every Azure resource belongs to exactly one subscription, and the subscription is what gets the bill. Subscriptions also carry limits and quotas, and they are a boundary for access control: giving someone access to a subscription gives them access to everything in it (subject to role). Organisations use multiple subscriptions to separate environments (production vs development), departments, or projects — for separate billing and separate access.

Resource groups are containers for resources within a subscription. A resource group holds the resources for one solution or workload — a web app, its database, its storage — so they can be managed, monitored, and deleted as a unit. The rules matter for the exam: a resource belongs to exactly one resource group; resource groups cannot be nested; deleting a resource group deletes everything in it; and the resources in a group can live in different regions (the group itself has a location, but its resources do not have to share it). Moving resources between groups is possible, with some restrictions.

Resources are the individual Azure services themselves — a virtual machine, a storage account, a virtual network, a database. They are the bottom of the hierarchy and the things you actually use.

Worked example: organising a company

A company with offices in Europe and Asia runs production and development environments. Sensible organisation: a management group per region (Europe, Asia) for governance, subscriptions per environment inside each (production, development) for separate billing and access, and resource groups per application inside each subscription (web shop, reporting, HR system). A policy applied at the Europe management group — "resources may only be created in European regions" — flows down to every subscription and resource group beneath it. A developer given access to the development subscription cannot touch production. And when the reporting project ends, deleting its resource group removes every resource cleanly. Hierarchy, billing, access, and lifecycle — one structure does all four.

Advertisement

Exam traps to watch for

Practice questions

Test yourself on this guide. Answers are checked instantly, with a full explanation for each question.

Q1.A company needs its virtual machines to keep running even if an entire Azure datacenter fails. Which feature should it use?

Report an error

Q2.What is the top level of the Azure resource hierarchy, used to apply governance across multiple subscriptions?

Report an error

Q3.A company wants its production and development environments billed separately, with developers unable to access production resources. What is the best way to organise this in Azure?

Report an error

Q4.Which statement about resource groups is correct?

Report an error

Q5.Azure replicates geo-redundant storage to a second region, and rolls platform updates out to one region at a time. What is the second region called?

Report an error

Q6.A company runs two virtual machines in a single availability set. What availability SLA does Azure offer for this configuration?

Report an error

Q7.A government agency must use an isolated instance of Azure that is physically and logically separated from the public cloud. Which Azure offering meets this need?

Report an error

Q8.Which of the following correctly lists the Azure resource hierarchy from the highest level to the lowest?

Report an error