Introduction to Microsoft Azure
What Microsoft Azure is, how subscriptions and resource groups organize it, its core services, and how to create your first resources safely.
Microsoft Azure is one of the three large public clouds, alongside AWS and Google Cloud. Most of its individual services look familiar if you know the other two: virtual machines, object storage, managed databases, Kubernetes. What makes Azure feel different is how it is organized, with every resource sitting inside a resource group, inside a subscription, inside a directory of identities that may already run your company’s email and laptops.
This article gives you the map. It covers where Azure fits, how its account structure works, which services make up the core, how the current names line up with the older ones you will still see in tutorials, and how to create and clean up your first resources from the command line. The rest of the Diving into Microsoft Azure series goes deeper on each piece.
Where Azure fits
Azure is Microsoft’s cloud platform for infrastructure, platform services and AI. It runs in regions around the world, most of which are split into availability zones: separate datacenters with their own power, cooling and networking. Azure Global Infrastructure covers regions, zones and region pairs in detail.
Technically, Azure covers the same ground as AWS and Google Cloud. The practical reasons teams choose it tend to come from the rest of Microsoft:
- Identity is already there. If your organization uses Microsoft 365, its users and groups live in Microsoft Entra ID, the same directory Azure uses for sign-in and access control. There is no second identity system to connect.
- Windows Server and SQL Server licensing. Azure Hybrid Benefit lets eligible customers apply their existing Windows Server and SQL Server licenses to Azure, which can change the cost comparison for Microsoft-heavy workloads.
- Hybrid management. Azure Arc projects servers and Kubernetes clusters running elsewhere, on premises or in another cloud, into Azure so you can apply the same policies and monitoring to them.
None of this makes Azure the right answer for everyone. AWS vs. GCP vs. Azure: Core Differences compares the three directly, and Choosing a Cloud Provider covers the decision itself.
How Azure is organized
Everything you create in Azure, whether a VM, a storage account or a database, is a resource. Resources are arranged in a strict hierarchy, and most of Azure’s access control, policy and billing hangs off it.
flowchart TD
T["Microsoft Entra tenant"] --> R["Root management group"]
R --> MG1["Management group: Production"]
R --> MG2["Management group: Sandbox"]
MG1 --> S1["Subscription: prod-web"]
MG2 --> S2["Subscription: dev-team"]
S1 --> RG1["Resource group: rg-web-prod"]
S1 --> RG2["Resource group: rg-data-prod"]
S2 --> RG3["Resource group: rg-experiments"]
RG1 --> V1["VM"]
RG1 --> V2["Load balancer"]
RG2 --> V3["SQL database"]
RG3 --> V4["Storage account"]
From the top down:
- Tenant. A dedicated instance of Microsoft Entra ID that holds your users, groups and app identities. Every subscription trusts exactly one tenant for sign-in.
- Management groups. Optional folders for subscriptions. Policies and role assignments set on a management group apply to everything below it. A single person with one subscription can ignore them; an organization with dozens of subscriptions relies on them.
- Subscriptions. The unit of billing and the boundary for many service quotas. A free account gives you one subscription; companies often use separate subscriptions for production, development and each team.
- Resource groups. Containers for resources that share a lifecycle. Each resource belongs to exactly one resource group, and groups can’t be nested.
- Resources. The things you actually pay for and use.
Resource groups are the part beginners most often get wrong, so a few rules are worth learning early. A resource group has a region, but that only sets where its metadata is stored; the resources inside it can be in any region. Deleting a resource group deletes every resource in it, which makes groups the cleanest way to tear down an experiment. And because permissions and policies are inherited, granting someone Contributor on a resource group lets them change everything inside it and nothing outside it.
A good default is one resource group per application per environment, such as rg-shop-dev and rg-shop-prod. Things that are created and deleted together belong together.
Azure Resource Manager
Every request to create, change or delete a resource goes through Azure Resource Manager (ARM), whichever tool sends it. The portal, the Azure CLI, Azure PowerShell, the SDKs, Bicep and ARM templates, and Terraform all call the same API.
flowchart LR
P["Azure portal"] --> ARM["Azure Resource Manager"]
C["Azure CLI or PowerShell"] --> ARM
I["Bicep, ARM templates, Terraform"] --> ARM
SDK["SDKs and REST"] --> ARM
ARM --> A{"Authenticated and authorized?"}
A -->|No| X["Request rejected"]
A -->|Yes| PO{"Allowed by Azure Policy?"}
PO -->|No| X
PO -->|Yes| RP["Resource provider, such as Compute or Storage"]
RP --> RES["Resource created or changed"]
This single path is why access control is consistent in Azure. A role assignment or a policy applies the same way whether someone clicks in the portal or runs a script, and every operation lands in the same activity log. It also means you can start in the portal and move to code later without learning a different permission model. Navigating the Azure Portal covers the portal side.
You create a resource group in Canada Central. Where can the VMs inside it run?
Correct! A resource group's location holds its metadata, and resources inside it can be in other regions.
The core services
Azure has hundreds of services, but most applications are built from a small set in four areas. The table maps common needs to the Azure service and the nearest equivalent on AWS and Google Cloud, which helps if you already know one of those.
| Need | Azure | AWS | Google Cloud |
|---|---|---|---|
| Virtual machines | Azure Virtual Machines | EC2 | Compute Engine |
| Managed Kubernetes | Azure Kubernetes Service (AKS) | EKS | GKE |
| Serverless containers | Azure Container Apps | ECS on Fargate | Cloud Run |
| Functions | Azure Functions | Lambda | Cloud Run functions |
| Object storage | Azure Blob Storage | S3 | Cloud Storage |
| Block storage for VMs | Azure Managed Disks | EBS | Persistent Disk |
| Private network | Azure Virtual Network | VPC | VPC |
| Managed relational database | Azure SQL Database, Azure Database for PostgreSQL | RDS, Aurora | Cloud SQL, AlloyDB |
| Globally distributed NoSQL | Azure Cosmos DB | DynamoDB | Firestore, Spanner |
Compute
Azure Virtual Machines give you Windows or Linux servers where you manage the operating system. Virtual Machine Scale Sets run groups of identical VMs that scale with load. Understanding Azure Virtual Machines explains the size families, and Creating Your First Azure Virtual Machine walks through building one.
Above VMs, Azure offers several managed ways to run code:
- Azure App Service hosts web apps and APIs from code or a container, with deployment slots, TLS and scaling handled for you.
- Azure Container Apps runs containers without a cluster to manage and can scale to zero.
- Azure Kubernetes Service (AKS) gives you a managed Kubernetes control plane when you need the full Kubernetes API.
- Azure Functions runs event-driven code. For new apps, Microsoft recommends the Flex Consumption plan; the older Linux Consumption plan is scheduled for retirement in September 2028.
flowchart TD
Q["What are you running?"] --> E{"Short, event-driven code?"}
E -->|Yes| F["Azure Functions"]
E -->|No| W{"A web app or API from code?"}
W -->|Yes| AS["App Service"]
W -->|No| C{"Containers?"}
C -->|"Yes, and you need Kubernetes APIs"| AKS["AKS"]
C -->|"Yes, without managing a cluster"| ACA["Container Apps"]
C -->|"No, you need OS control"| VM["Virtual Machines"]
The trade-off runs in one direction: the further down the list you go, toward VMs, the more control you get and the more patching, scaling and monitoring you take on. Cloud Service Models: IaaS, PaaS, SaaS explains that spectrum in general terms.
Storage
Most Azure storage lives in a storage account, a named container with its own endpoint, redundancy setting and access keys. One account can hold several kinds of data:
- Blob Storage for unstructured objects such as backups, images, logs and data lake files, with hot, cool, cold and archive tiers.
- Azure Files for SMB and NFS file shares that VMs and on-premises machines can mount.
- Queue Storage and Table Storage for simple messaging and key-value data.
VM disks are separate: Managed Disks are block devices attached to a VM, sold in tiers from standard HDD to Premium SSD v2 and Ultra Disk.
The redundancy setting on a storage account matters more than people expect. Locally redundant storage keeps copies within one datacenter, zone-redundant storage spreads them across availability zones, and the geo-redundant options also copy data to a second region. Pick it deliberately, because the cheapest option doesn’t survive a zone outage.
Networking
An Azure Virtual Network (VNet) is your private address space in a region, divided into subnets. Network security groups filter traffic to subnets and network interfaces. In front of your apps you can place Azure Load Balancer for TCP and UDP traffic, Application Gateway for HTTP routing and a web application firewall within a region, or Azure Front Door for global HTTP entry with caching. Azure DNS hosts your domains.
Databases
Azure SQL Database is a managed SQL Server engine, and Azure SQL Managed Instance covers workloads that need near-complete SQL Server compatibility. Azure Database for PostgreSQL and Azure Database for MySQL run the open-source engines on the flexible server deployment model. Azure Cosmos DB is a globally distributed database with several APIs, including NoSQL, MongoDB and Apache Cassandra.
You have a containerized API, traffic that drops to nothing overnight, and no need for Kubernetes APIs. Which service fits best?
Correct! Container Apps runs containers without a cluster to manage and can scale to zero when idle.
Data, AI and security
Beyond the core, three areas come up in almost every Azure conversation.
Data and analytics. Microsoft Fabric is where Microsoft is putting its new analytics investment: a SaaS platform that combines data engineering, warehousing, real-time analytics and Power BI around a shared data lake. Azure Synapse Analytics is still supported, but new projects generally start on Fabric. Azure Data Factory moves and transforms data, and Azure Databricks runs the Databricks platform as a first-party Azure service.
AI. Microsoft Foundry is the platform for building AI applications and agents, with access to models from OpenAI and other providers. Foundry Tools are the prebuilt services for speech, vision, language, translation and document processing. Azure Machine Learning remains the service for training and operating your own models. Cloud Services for Generative AI: Azure Offerings covers this area in more depth.
Security and identity. Microsoft Entra ID handles sign-in, and Azure role-based access control (RBAC) decides what each identity can do at each level of the hierarchy. Azure Key Vault stores secrets, keys and certificates so they stay out of your code. Microsoft Defender for Cloud assesses your configuration against security recommendations and adds paid threat protection plans; check that its free posture management is turned on, because new subscriptions now start with it off. Microsoft Sentinel is Microsoft’s cloud SIEM; its management is moving into the Microsoft Defender portal, and Microsoft has set March 31, 2027 as the date it can no longer be managed from the Azure portal. Azure Identity and Access Management explains Entra ID and RBAC.
Old names you will still see
Azure renames products often, and older tutorials, including an earlier version of this article, use names that no longer appear in the portal. If a guide mentions one of these, look for its current name.
| Older name | Current name |
|---|---|
| Azure Active Directory (Azure AD) | Microsoft Entra ID |
| Azure Security Center, Azure Defender | Microsoft Defender for Cloud |
| Azure Sentinel | Microsoft Sentinel |
| Cognitive Services, Azure AI services | Foundry Tools |
| Azure AI Studio, Azure AI Foundry | Microsoft Foundry |
| Single Server for PostgreSQL and MySQL | Flexible server (Single Server is retired) |
A 2022 tutorial tells you to create a group in Azure Active Directory. Where do you do that today?
Correct! Azure Active Directory was renamed Microsoft Entra ID; groups, users and app registrations work the same way.
Getting started
New customers can start with an Azure free account, which includes a credit for the first 30 days plus free monthly amounts of selected services. Azure Free Tier and Billing Overview explains how the credit, the spending limit and the free services interact, and Setting Up Free Accounts on AWS, GCP, and Azure walks through sign-up. Before you build anything, set a budget with an email alert in Cost Management so a forgotten resource can’t surprise you.
The portal is the easiest place to look around, but the Azure CLI shows the structure more clearly. You can install it locally or open Cloud Shell from the portal, where it is already signed in. This sequence signs in, checks which subscription you are working in, creates a resource group, and deletes it again:
# Sign in (opens a browser; skip this in Cloud Shell)
az login
# Show the active subscription and the tenant it belongs to
az account show --query "{subscription:name, id:id, tenant:tenantId}" --output table
# List the regions available to your subscription
az account list-locations --query "[].name" --output tsv
# Create a resource group for experiments
az group create --name rg-intro-dev --location canadacentral
# See every resource group in the subscription
az group list --output table
# Delete the group and everything in it when you're done
az group delete --name rg-intro-dev --yes --no-wait
If you have access to more than one subscription, run az account set --subscription "<name or id>" before creating anything, so resources land where you expect them. Creating everything for an experiment in one resource group, and deleting that group afterwards, is the simplest habit for keeping costs and clutter down.
Further reading
- Azure documentation is the starting point for every service.
- What is Azure Resource Manager? explains scopes, resource groups and deployments.
- Azure Cloud Adoption Framework covers how organizations structure management groups and subscriptions.
- Azure Architecture Center has reference architectures and design guidance.
- Azure CLI documentation lists every command.
- Azure pricing calculator estimates costs before you deploy.
Key takeaways
- Azure organizes everything as tenant, management groups, subscriptions, resource groups and resources, and access control and policy are inherited down that tree.
- A resource group’s region only stores metadata; group resources by shared lifecycle, and delete the group to remove an experiment completely.
- Every portal click, CLI command and infrastructure-as-code deployment goes through Azure Resource Manager, so permissions and logs are consistent across tools.
- The core services map closely to AWS and Google Cloud equivalents; choose compute by how much of the platform you want to manage.
- Many tutorials use retired names: Azure AD is Microsoft Entra ID, Security Center is Defender for Cloud, and Cognitive Services are now Foundry Tools.
- Set a budget alert before you create resources, and check which subscription is active before every deployment.
Part 1 of 7 · Diving into Microsoft Azure
- 01Introduction to Microsoft Azure
- 02Navigating the Azure Portal
- 03Azure Global Infrastructure: Regions and Availability Zones
- 04Mastering Microsoft Entra Identity and Access Management
- 05Azure Free Tier and Billing Overview
- 06Creating Your First Azure Virtual Machine
- 07Azure Support and Documentation Resources