Skip to main content

Optimization

Implement cost optimization strategies to reduce cloud spending without sacrificing performance.

Optimization Strategies​

Reserved Instances​

Convert On-Demand instances to Reserved Instances for 30-60% savings:

  • Standard Reserved Instances: Highest discount, fixed instance type
  • Convertible Reserved Instances: Lower discount, flexible instance types
  • 1-Year or 3-Year Terms: Longer commitment = higher savings

Best For: Stable, predictable workloads running 24/7

Savings Plans​

Flexible pricing model based on compute usage commitment:

  • Compute Savings Plans: Most flexible, any instance family
  • EC2 Instance Savings Plans: Higher discount, specific instance family
  • SageMaker Savings Plans: For ML workloads

Best For: Workloads with variable instance types but consistent compute usage

Right-Sizing​

Downsize overprovisioned resources:

  • Analyze CPU, memory, and disk utilization
  • Identify instances with < 20% average utilization
  • Recommend smaller instance types
  • Test and validate before applying changes

Potential Savings: 20-40% on compute costs

Spot Instances​

Use spare EC2 capacity at up to 90% discount:

  • Best For: Batch jobs, data processing, fault-tolerant workloads
  • Not Recommended For: Production databases, stateful applications
  • Combine with On-Demand for mixed fleet strategy

Auto-Scaling​

Automatically adjust capacity based on demand:

  • Scale out during high traffic periods
  • Scale in during low traffic periods
  • Set minimum and maximum instance counts
  • Use metrics-based or schedule-based scaling

Potential Savings: 30-50% by matching capacity to demand

Storage Optimization​

Reduce storage costs through lifecycle management:

  • S3 Intelligent Tiering: Automatically move objects between access tiers
  • Glacier Archive: Move old data to low-cost archive storage (90% savings)
  • EBS Snapshot Cleanup: Delete old or unused snapshots
  • Unattached Volume Cleanup: Remove volumes not attached to instances

Automated Scheduling​

Stop non-production resources during off-hours:

Development Environment Schedule​

  • Weekdays: 8 AM - 7 PM (11 hours/day)
  • Weekends: Stopped
  • Savings: 60-70% on development compute costs

Staging Environment Schedule​

  • Business Hours: 6 AM - 10 PM (16 hours/day)
  • Weekends: Limited hours or stopped
  • Savings: 30-40% on staging compute costs

Implementation​

  1. Tag resources with environment:dev or environment:staging
  2. Configure schedule in FinOps → Settings → Automated Scheduling
  3. Exclude critical resources with cost-optimization:exclude tag
  4. Monitor savings in FinOps dashboard

Continuous Optimization​

Best practices for ongoing cost management:

  • Weekly Reviews: Check cost trends and anomalies
  • Monthly Optimization: Apply new recommendations
  • Quarterly Analysis: Review Reserved Instance coverage
  • Annual Planning: Adjust budgets and optimization strategies
Best Practice

Start with quick wins (unused resources, automated scheduling) for immediate savings, then implement long-term strategies (Reserved Instances, Savings Plans) for sustained cost reduction.