Every architecture decision in Azure must balance four pillars:
1. Security
-
Protect identity → MFA, Conditional Access, PIM.
-
Protect data → Encryption at rest (TDE, ADE, SSE), Encryption in transit (TLS/VPN).
-
Protect secrets → Key Vault.
-
Protect apps → App Gateway WAF, DDoS.
2. Scalability
-
Scale up → Bigger VM size.
-
Scale out → VM Scale Sets, AKS pods, App Service instances.
-
Auto-scale → Based on CPU, memory, or demand metrics.
3. Reliability & High Availability
-
Availability Sets (rack-level resilience).
-
Availability Zones (datacenter-level resilience).
-
Geo-redundancy (cross-region failover).
-
Load balancing (L4 vs L7).
-
Disaster Recovery (ASR).
4. Cost Optimization
-
Use reserved instances for steady workloads.
-
Spot VMs for fault-tolerant jobs.
-
Right-size storage tiers (Hot, Cool, Archive).
-
Set budgets and alerts.
Confusion Buster 🚨
-
Don’t confuse availability (system stays up) with disaster recovery (system recovers after going down).
-
Don’t assume “best design” always means most expensive — exam often expects you to optimize costs while meeting requirements.
Example Scenario
A startup runs an app that:
-
Must scale during flash sales.
-
Needs 99.99% uptime.
-
Has a limited budget.
➡️ Correct Design:
-
Deploy web tier in VM Scale Sets across Availability Zones.
-
Use App Gateway with WAF for front-end.
-
Enable auto-scaling rules.
-
Use Spot VMs for background jobs.
Exam Tip
When two answers look correct, pick the one that:
-
Meets requirements (security, uptime, compliance).
-
Is cost-optimized.
-
Uses PaaS/serverless over IaaS if possible.
What to Expect in the Exam
-
“Company needs 99.99% uptime for VMs.” → Availability Zones.
-
“Which principle ensures system continues to function during hardware failure?” → Reliability/HA.
-
“How to reduce cost of steady-state workloads?” → Reserved Instances.