AZ-900 Exam Review: Key Concepts Summary
Key Takeaways
- AZ-900 logistics: about 40-60 questions in 45 minutes, scored 0-1000, pass at 700, USD $99 voucher, no prerequisites, valid permanently (no renewal).
- Domain 1 Cloud Concepts (25-30%): IaaS/PaaS/SaaS, shared responsibility, CapEx vs OpEx, public/private/hybrid, scalability vs elasticity.
- Domain 2 Architecture & Services (35-40%): regions, availability zones, region pairs, the management group > subscription > resource group > resource hierarchy, and core compute/network/storage/identity services.
- Domain 3 Management & Governance (30-35%): Pricing/TCO calculators, Cost Management, Azure Policy vs RBAC, locks, tags, ARM/Bicep, Monitor, and SLA math.
- The exam rewards breadth: know WHAT each service does and WHEN to use it, eliminate wrong options, and never leave a question blank.
Exam logistics you should walk in knowing (verified 2026)
| Item | Fact |
|---|---|
| Exam code / name | AZ-900: Microsoft Azure Fundamentals |
| Questions | Roughly 40-60 items (form varies); mix of multiple-choice, multi-select, drag-and-drop, and dropdowns |
| Time | About 45 minutes of testing |
| Scoring | Scaled 0-1000; pass at 700 (not a raw 70%) |
| Cost | USD $99 voucher (varies by country/taxes) |
| Prerequisites | None |
| Validity | Fundamentals certifications do not expire — no annual renewal |
| Objectives updated | January 14, 2026 (expanded AI, Copilot, Responsible AI) |
Domain 1: Cloud Concepts (25-30%)
| Concept | Key fact |
|---|---|
| CapEx vs OpEx | CapEx = buy hardware up front; OpEx = pay-as-you-go for cloud services |
| IaaS | You manage OS + runtime + apps + data (Azure VMs) |
| PaaS | Provider manages OS; you manage app + data (App Service) |
| SaaS | You manage only data + settings (Microsoft 365) |
| Shared responsibility | The customer always owns data, accounts/identities, and devices, regardless of model |
| Public / Private / Hybrid | Multi-tenant / single-tenant dedicated / a mix of both |
| Scalability vs Elasticity | Scalability = ability to grow; elasticity = automatic scale up and down with demand |
| High availability | Redundancy that keeps a service running, backed by an SLA |
Trap: Scalability and elasticity are different — elasticity is the automatic one. And no matter how much the provider manages, the customer never offloads responsibility for data and identities.
Domain 2: Azure Architecture & Services (35-40%)
Global infrastructure & hierarchy
| Concept | Key fact |
|---|---|
| Region | A set of data centers in one geography (60+ worldwide) |
| Availability Zone | 3+ physically separate data centers in a region; survives a single DC failure |
| Region pair | Two regions 300+ miles apart; staggered updates for disaster recovery |
| Hierarchy | Management group > Subscription > Resource group > Resource |
| Resource group | Logical container; each resource lives in exactly one group |
| ARM | Azure Resource Manager — the single control plane for all management tools |
Compute & networking at a glance
| Service | When to use |
|---|---|
| Azure VMs (IaaS) | Full OS control, lift-and-shift |
| App Service (PaaS) | Web apps and APIs without managing servers |
| Azure Functions | Event-driven, serverless, pay-per-execution |
| AKS | Orchestrate complex multi-container microservices |
| VPN Gateway | Encrypted tunnel over the internet to on-prem |
| ExpressRoute | Private circuit, not over the internet, up to 100 Gbps |
| Load Balancer / App Gateway / Front Door | L4 / L7+WAF / global L7+CDN |
Storage & identity
| Service | Key fact |
|---|---|
| Blob / Files / Queue / Table | Unstructured / SMB shares / async messaging / NoSQL key-value |
| LRS, ZRS, GRS, GZRS | 3 copies in one DC, across zones, across regions, or zones+regions |
| Entra ID | Cloud identity: SSO, MFA, Conditional Access |
| RBAC vs Azure Policy | RBAC = who can do things; Policy = what resources are allowed |
Domain 3: Management & Governance (30-35%)
| Concept | Key fact |
|---|---|
| Pricing Calculator | Estimate cost before deploying |
| TCO Calculator | Compare on-premises vs Azure cost |
| Cost Management | Track actual spend, set budgets and alerts |
| Azure Advisor | Free recommendations across cost, security, reliability, performance, operations |
| Azure Policy | Enforce rules (e.g., allowed regions, required tags) |
| Resource locks | CanNotDelete or ReadOnly to prevent accidents |
| Tags | Key-value metadata; not inherited by child resources by default |
| ARM templates / Bicep | JSON IaC / simpler DSL that compiles to ARM JSON |
| Azure Monitor / Log Analytics / App Insights | Telemetry / KQL log queries / app performance monitoring |
| Service Health | Personalized Azure outage and maintenance notices |
SLA math you must be able to do
A 99.9% monthly SLA allows about 43 minutes of downtime per month. When you chain dependent services, you multiply their SLAs, so the composite SLA always drops: 99.9% × 99.9% = 99.81%. Adding more dependencies lowers it further — a frequent exam trap.
Final exam-day checklist
- Three service models (IaaS/PaaS/SaaS) and the shared responsibility split
- Public / private / hybrid deployment models
- Regions, Availability Zones, region pairs
- The four-level resource hierarchy
- VMs vs App Service vs Functions
- VPN Gateway vs ExpressRoute
- Storage redundancy (LRS/ZRS/GRS/GZRS) and tiers
- Entra ID, RBAC vs Azure Policy, Zero Trust principles
- Pricing vs TCO Calculator vs Cost Management
- Locks, tags, ARM/Bicep, Monitor, and SLA multiplication
Final Tip: AZ-900 tests breadth, not depth — you do not configure or code anything. Pass at 700/1000, manage your 45 minutes, eliminate clearly wrong options, trust your first read, and never leave a question blank.
What is the passing score for the AZ-900 exam?
Two services in an architecture each have a 99.9% SLA and one depends on the other. What is the composite SLA?
A company must protect data against a complete Azure region failure. Which storage redundancy option meets the requirement?
Which THREE statements are principles of the Zero Trust model? (Select THREE)
Select all that apply