Why Optimization Matters
After workloads are migrated to Azure, many organizations face bill shock or performance bottlenecks because resources are not tuned for the cloud. The exam expects you to know how to optimize both cost and performance while ensuring workloads still meet SLA and compliance requirements.
1. Cost Optimization Strategies
Azure Advisor
-
Provides best-practice recommendations for cost, performance, reliability, and security.
-
Example: Identify underutilized VMs that can be resized or shut down.
Reserved Instances (RI)
-
1-year or 3-year commitment for VMs, SQL DB, Cosmos DB.
-
Up to 72% cheaper than pay-as-you-go.
Spot VMs
-
Deeply discounted VMs (up to 90%) but can be evicted if capacity is needed.
-
Good for batch jobs or dev/test workloads.
Azure Hybrid Benefit
-
Reuse existing Windows Server & SQL Server licenses in Azure.
-
Saves significant costs for enterprise workloads.
Right-Sizing
-
Monitor VM CPU/memory trends → scale down oversized machines.
Auto-Shutdown & Scheduling
-
Shut down dev/test VMs during off-hours.
Storage Tiering
-
Hot tier = frequently accessed.
-
Cool/Archive tier = infrequently accessed data.
2. Performance Optimization Strategies
Autoscaling
-
Scale out/in based on demand (VMSS, App Service Plans).
-
Prevents overprovisioning while ensuring SLA.
Caching
-
Azure Cache for Redis → reduce DB load, faster response.
-
Content Delivery Network (CDN) → reduce latency for global users.
Database Tuning
-
Use SQL Advisor for index optimization.
-
Enable geo-replication for better performance near users.
Network Optimization
-
ExpressRoute for low-latency hybrid connectivity.
-
Azure Front Door / Traffic Manager for global routing.
Monitoring Bottlenecks
-
Use Application Insights for dependency tracking.
-
Use Log Analytics to identify underperforming resources.
Example Enterprise Scenario
A SaaS company faces:
-
Azure bill unexpectedly high.
-
Customer portal slow in APAC region.
-
SQL queries taking too long during peak hours.
Correct design:
-
Apply Azure Advisor → right-size VMs, shut down dev workloads after hours.
-
Move customer portal content to CDN + Front Door for APAC acceleration.
-
Enable SQL index tuning + Redis Cache for query performance.
Confusion Buster
-
Reserved Instance vs Spot VM
-
RI = predictable workloads, guaranteed capacity.
-
Spot = variable workloads, cheaper but no SLA.
-
-
Hybrid Benefit vs Reserved Instance
-
Hybrid = saves by reusing licenses.
-
RI = saves via commitment to Azure.
-
-
Autoscale vs Manual Scaling
-
Autoscale = automatic resource adjustment.
-
Manual = fixed size, risk of over/under provision.
-
Exam Tips
-
“Which Azure service provides recommendations for cost optimization?” → Azure Advisor.
-
“Which option is best for long-term predictable workloads?” → Reserved Instances.
-
“Which VM type is cheapest but unreliable for SLA workloads?” → Spot VM.
-
“Which feature allows reusing existing Windows/SQL licenses?” → Hybrid Benefit.
-
“Which service caches data to improve performance?” → Azure Cache for Redis.
What to Expect in the Exam
-
Direct Q: “Which Azure service suggests resizing VMs to save cost?” → Azure Advisor.
-
Scenario Q: “Company needs low-cost compute for batch jobs with no SLA.” → Spot VMs.
-
Scenario Q: “Company wants to optimize SQL performance with caching.” → Redis Cache.
-
Scenario Q: “Company has licenses for Windows Server on-prem and wants to save on Azure.” → Hybrid Benefit.
-
Trick Q: “Spot VMs are guaranteed not to be evicted.” → False.