🧠 Compute Demystified: Powering the Cloud, One VM at a Time

Hey there, cloud cadet!
I'm currently preparing for the AZ-900: Microsoft Azure Fundamentals exam, and as part of my learning journey, I’ve decided to share my notes with you in a simple, no-fluff format.

Today, we’re diving into one of the core building blocks of the cloud: Compute services.

✅ Core Azure Compute Services

💻 Azure Virtual Machines (VMs)IaaS

  • You manage: OS, apps, patches, security.

  • Great for legacy apps, custom software, or lift-and-shift scenarios.

  • Pricing based on VM size, OS, storage, and region.

  • Supports availability sets, zones, and scale sets.

VM Series (What to Remember):

  • B-series: Burstable workloads.

  • D-series: General purpose (web servers, databases).

  • E-series: Memory optimized (SQL, cache).

  • F-series: Compute optimized (CPU-heavy).

  • Lsv2-series: Storage optimized (high IOPS).

  • N-series: GPU (AI, ML, rendering).

  • M-series: High memory (SAP).

🌐 Azure App ServicePaaS

  • Host web apps and APIs.

  • Supports .NET, Node.js, Java, Python, PHP, and more.

  • App Service Plan determines:

    • Region

    • Pricing tier

    • Scaling (manual/auto)

  • Multiple apps can share the same App Service Plan.

App Service Plan Tiers:

  • Free/Shared – Dev/test only

  • Basic – Manual scale

  • Standard – Auto-scale + staging

  • Premium – Faster & more scalable

  • Isolated – Secure, in VNet

📦 Azure Container Instances (ACI)

  • Run containers without managing servers.

  • No orchestration (single container or simple groups).

  • Best for quick tasks, jobs, or APIs.

  • Pay per second of CPU/memory used.

🚢 Azure Kubernetes Service (AKS)

  • Managed Kubernetes to run containerized microservices.

  • Azure manages the master nodes.

  • Supports auto-scaling, updates, and RBAC.

  • Best for complex, production-grade container workloads.

Azure FunctionsServerless

  • Event-driven code execution (e.g. timers, HTTP triggers).

  • Scales automatically based on events.

  • Billing: Pay only for executions and runtime.

Max Execution Time:

  • Consumption plan: 5 minutes (extendable to 10)

  • Premium/Dedicated: Unlimited

🧑‍💻 Azure Virtual Desktop (AVD)

  • Host Windows desktops in the cloud.

  • Accessible from anywhere.

  • Integrates with Microsoft 365 and Azure AD.

  • Useful for secure remote access or centralized desktops.

🧮 Azure Batch

  • Run large-scale batch or parallel jobs.

  • Automatically provisions and manages compute.

  • Ideal for rendering, modeling, simulations.

  • Works with custom scripts, containers, and apps.

🛡️ High Availability & SLAs

Service

SLA

VM + Availability Set

99.95%

VM + Availability Zones

99.99%

App Service (Standard+)

99.95%

Functions (Premium Plan)

99.95%

AKS Nodes

99.95%

🧠 Free tiers don’t have an SLA!

📊 Scaling Options

Service

Scaling

VMs

Manual or auto (with Scale Sets)

App Service

Manual or auto (Standard+)

AKS

Auto-scaling with node pools

Functions

Auto-scale (by default)

ACI

Manual per container group

💰 Compute Pricing Models

  • VMs: Per second; price varies by region and specs.

  • App Service: Based on App Service Plan.

  • ACI / Functions: Per execution (true pay-per-use).

  • AVD: Per user and VM usage.

  • AKS: Pay for agent nodes (control plane is free).

  • Batch: Depends on the VM resources used.

🔐 Security & Identity

  • Use Azure AD + RBAC to control access to compute resources.

  • Apply NSGs (Network Security Groups) to protect VMs.

  • App Service and Functions support managed identities.

🌍 Hybrid & Management Tools

  • Azure Arc: Manage on-prem VMs & Kubernetes like Azure resources.

  • Management Tools:

    • Azure Portal (GUI)

    • Azure CLI / PowerShell

    • ARM Templates or Bicep (Infrastructure as Code)

📌 Exam Tips (AZ-900 Specific)

  • Know the difference between IaaS, PaaS, Serverless.

  • Expect scenario-based questions: “Which service should you choose if…”

  • Memorize SLA levels and scaling capabilities.

  • Understand when to use each compute service and its pricing implications.