Microsoft Azure
Compact study notes on Microsoft Azure: core concepts, services, architecture, management, security, SLAs and common exam-style problems for B.E./B.Tech students.
Drafted with Aria, reviewed by the AiCanCode.org team · Updated . How we write and check our notes. Spotted an error? Use Give Feedback at the bottom of the page.
Why it matters
Microsoft Azure is a major cloud platform used in industry. Engineers use Azure to host applications, store and process data, provide scalable compute and networking, and implement enterprise identity and security — so students must know Azure concepts, common services, how to design for availability and cost, and basic deployment commands.
Key ideas
Before using a term, its meaning is defined in simple words.
Cloud platform: a set of remote compute, storage and networking services delivered over the internet. Assumes basic networking and virtualization exist.
Azure: Microsoft’s public cloud platform that offers IaaS, PaaS and SaaS services over a global network of datacentres. (Definition: an integrated set of managed services you can consume on demand.)
Subscription: a billing and access boundary associated with an Azure account. Each subscription is a container for resources and has a subscription ID.
Resource group: a logical container that holds related Azure resources (VMs, networks, storage accounts) for management and access control. All resources in a group share lifecycle operations (create, delete).
Region: a geographical location (for example, 'Central India') where Azure has datacentres. A region provides physical locality and latency considerations.
Availability Zone: an isolated location within a region with independent power, cooling and networking. Availability Zones improve fault tolerance. (Assumption: a region supports zones only if Microsoft advertises them for that region.)
Azure Resource Manager (ARM): the control plane and deployment system for Azure resources. ARM accepts declarative templates (ARM templates / Bicep) to create and manage resources.
IaaS (Infrastructure as a Service): cloud model where compute and storage are provided as virtual machines/disks; you manage OS and applications. In Azure: Virtual Machines (VMs) and Managed Disks.
PaaS (Platform as a Service): cloud model where the provider manages OS and runtime; you deploy applications. In Azure: App Service, Azure SQL Database, Azure Functions.
SaaS (Software as a Service): complete software delivered over the internet (e.g., Office 365). The provider manages everything.
Virtual Machine (VM): a virtualized server instance with virtual CPU, memory, storage and networking. In Azure you choose VM SKU, size and OS image.
VM Scale Set: a group of identical VMs managed as a set that can autoscale. Use for stateless scalable workloads.
Availability Set: a logical grouping of VMs to spread them across fault domains and update domains within a single data center in a region to reduce simultaneous outages.
Managed Disk: durable block storage attached to VMs. Microsoft manages the storage account backing the disks.
Blob Storage: object storage for unstructured data (files, images, backups). It has access tiers (Hot, Cool, Archive) for cost optimization. Redundancy options include LRS, ZRS, GRS, and RA-GRS; choose based on durability and regional resilience requirements.
Azure Files: fully managed file shares accessible using SMB/NFS.
Azure Cosmos DB: globally distributed, multi-model NoSQL database with guaranteed low latency and configurable consistency.
Azure SQL Database: managed relational database (PaaS) compatible with Microsoft SQL Server. Access it securely via Private Link (Private Endpoint) or service endpoints.
Virtual Network (VNet): an Azure network boundary that provides IP addressing, subnets, routing and isolation for resources.
Network Security Group (NSG): a set of firewall rules applied to subnets or network interfaces to allow or deny traffic.
Load Balancer: distributes incoming traffic among backend VMs at layer 4 (TCP/UDP). For layer 7, Azure provides Application Gateway.
Microsoft Entra ID (formerly Azure Active Directory/Azure AD): Microsoft’s cloud identity and access management service. It provides authentication, single sign-on, and directory services.
Role-Based Access Control (RBAC): a permission model where roles (e.g., Reader, Contributor, Owner) are assigned to users/groups at scopes (subscription, resource group, resource).
Azure Key Vault: secure storage for secrets, keys and certificates. Use with Managed Identities to eliminate embedded credentials in code.
Azure Monitor and Log Analytics: services for collecting metrics, logs and alerts for resources and applications.
Azure Policy: service to enforce rules and standards across resources (for example, allowed VM sizes or regions).
ExpressRoute: private connection between on-premises infrastructure and Azure (not over public internet). Use for predictable bandwidth and more consistent, typically lower, latency than the public internet.
Azure Functions: serverless compute service that runs short-lived functions triggered by events. Billed per execution and resource consumption.
Azure Kubernetes Service (AKS): managed Kubernetes control plane; you manage worker nodes or use virtual node integrations.
Definitions that are often tested (exact statements):
- Definition (SLA): A Service Level Agreement (SLA) is a formal document from the provider that declares the guaranteed uptime percentage for a service under specified conditions. Assumptions: measurement method and service configuration must match the SLA’s stated prerequisites.
- Definition (High availability): A system is highly available if it remains operational for a large fraction of the time by design, using redundancy and fault isolation.
Diagram instructions often requested in exams:
- Draw a simple Azure deployment diagram for a web app: draw a box labelled 'Region', inside it draw VNet, inside VNet draw two subnets: 'frontend subnet' with VMs or AKS nodes behind an Application Gateway; 'backend subnet' with a Private Endpoint to Azure SQL Database (PaaS). If using App Service (multitenant), show it outside the VNet with Private Endpoint/VNet integration as needed. Show arrows: public internet -> Application Gateway (L7) -> Backend VMs/VMSS (no extra Load Balancer needed), or alternatively public internet -> Load Balancer (L4) -> Backend VMs. Label Availability Zones or an Availability Set used for backend VMs.
Formulas
A_total = 1 - Π_i (1 - A_i) — Combined availability of parallel independent components.
- A_i: availability (as fraction, e.g., 0.999) of component i. No units. Applies when components fail independently and the system is up if at least one component is up.
Downtime = (1 - SLA) × Period — Expected downtime in the given period.
- SLA: availability fraction (e.g., 0.9995 for 99.95%).
- Period: total time (use same units as desired downtime output, e.g., hours/year = 365×24 hours).
Required_instances = ceil( log(1 - Target_A) / log(1 - A_single) ) — Minimum identical independent instances in parallel to reach a target availability, derived from A_total formula rearranged for identical A_i = A_single. Applies under independence assumption.
(Commands to remember — syntax examples)
az group create --name MyRG --location centralindia — create resource group.
az vm create --resource-group MyRG --name MyVM --image UbuntuLTS --admin-username azureuser --generate-ssh-keys — create a VM with SSH keys.
az role assignment create --assignee user@example.com --role Reader --scope /subscriptions/{subId}/resourceGroups/MyRG — assign RBAC role.
az deployment group create --resource-group MyRG --template-file template.json --parameters @params.json — deploy ARM template.
(Verify these command syntaxes against the latest Azure CLI version; commands may change between versions.)
Worked examples
Example 1 — SLA downtime calculation (typical exam question) Problem: Azure guarantees an SLA of 99.95% for a given service. Calculate the maximum downtime per year and per month.
Step 1: Write formula: Downtime = (1 - SLA) × Period.
Step 2: Compute per year (use 365 days): Period_year = 365 × 24 = 8760 hours.
Downtime_year = (1 - 0.9995) × 8760 = 0.0005 × 8760 = 4.38 hours/year.
Step 3: Compute per 30-day month: Period_month = 30 × 24 = 720 hours.
Downtime_month = 0.0005 × 720 = 0.36 hours = 21.6 minutes/month.
Answer: ~4.38 hours/year and ~21.6 minutes/month.
Example 2 — Designing for target availability with identical VMs Problem: Each VM has availability 99.5% (A_single = 0.995). Find the minimum number of identical VMs in parallel (with load balancing) required to reach overall availability 99.999% (Target_A = 0.99999). Assume independent failures.
Step 1: Use formula for n identical instances: A_total = 1 - (1 - A_single)^n.
Step 2: Rearranged: (1 - A_single)^n = 1 - A_total.
n = log(1 - A_total) / log(1 - A_single).
Compute: 1 - A_total = 1 - 0.99999 = 0.00001.
1 - A_single = 1 - 0.995 = 0.005.
n = log(0.00001) / log(0.005).
Numeric: log10 approach: log10(0.00001) = -5. log10(0.005) ≈ -2.30103.
n ≈ (-5)/(-2.30103) ≈ 2.171.
Take ceiling: n = 3.
Answer: 3 identical VMs (with independent failure behavior and proper load balancing) are required to meet 99.999%.
Example 3 — Quick deploy a VM (command trace, typical lab question) Problem: Create a resource group and a Ubuntu VM using Azure CLI; list the commands and explain each step.
Step 1: Create resource group:
az group create --name StudentRG --location centralindia — creates a resource group called StudentRG in Central India region.
Step 2: Create VM:
az vm create --resource-group StudentRG --name ExamVM --image UbuntuLTS --admin-username azureuser --generate-ssh-keys — creates VM named ExamVM using Ubuntu image, generates SSH keys, sets username.
Step 3: Open port 22 (if not opened automatically):
az vm open-port --resource-group StudentRG --name ExamVM --port 22 — creates an NSG rule to allow SSH.
Explain: After these commands, VM is reachable by SSH using the private key generated locally.
(Commands are dependent on Azure CLI version and RBAC permissions of your account.)
Common mistakes
- Treating Azure AD as the same as on-premises Active Directory without mentioning federation or Azure AD Connect for hybrid identity.
- Forgetting constraints of SLAs: many SLAs require configuration (e.g., two or more instances behind a load balancer) to be valid.
- Misreading storage tiers: Archive tier is for rarely accessed data and has retrieval latency and costs — not for primary active data.
- Assuming Azure Functions are free to run unlimited: they are billed per execution and resource time; pricing model matters for long-running processes.
- Not validating ARM template schema versions or Bicep-to-ARM compilation before deployment.
For your exam
Question types you will get from this topic:
- Definitions and short notes: define Azure, resource group, VM, VNet, Azure AD, ARM. Full-marks include concise definitions and one example use.
- Diagrams: draw an Azure deployment with region, VNet, subnets, load balancer and backend resources. Label zones or sets. Full-marks: correctly show resource group boundaries and the public internet arrow.
- Derivations and formulas: derive combined availability of parallel components; calculate downtime from SLA. Full-marks: show formula, substitute values, and carry units.
- Configuration/commands: show Azure CLI commands to create resource group, VM, assign roles, or deploy ARM template. Full-marks: correct commands and explanation of flags and prerequisites (e.g., logged-in user, role permissions).
- Short lab-style trace: given a small ARM template, explain resource creation order and dependencies. Full-marks: identify implicit/explicit dependencies and explain outputs.
Exam tips:
- Always state assumptions (independence, number of instances, measurement period) when calculating availability.
- When asked about security, mention Azure AD, RBAC, Key Vault, NSGs and Azure Policy together to show layered approach.
Quick check
- What is a Resource Group in Azure?
- Write the formula to compute expected downtime from SLA for one year.
- Which Azure service would you use to store secrets and keys securely?
Answers: 1) A logical container for related resources used for management and lifecycle. 2) Downtime = (1 - SLA) × 8760 hours. 3) Azure Key Vault.
Finished this topic? Mark it so your progress, study plan and readiness keep up.
Stuck on something here?