How do you optimize cloud infrastructure costs? What are common cost reduction strategies?

Core Principles of Cloud Cost Optimization

Right-sizing: Choose instance types that best match your workload — avoid over-provisioning.

Eliminate waste: Find and delete unused resources (orphaned EBS volumes, unused Elastic IPs, old snapshots).

Purchase strategy optimization: Choose the right payment model based on workload stability.

Payment Model Comparison

Model	Discount	Best For
On-Demand	No discount	Short-term, unpredictable workloads
Reserved Instances (1-3 yr)	Up to 72%	Stable production baseline load
Savings Plans	Up to 66%	Flexible compute commitment
Spot Instances	Up to 90%	Fault-tolerant batch jobs, CI/CD tasks

Common Optimization Strategies

Compute layer

• Use Spot Instances for non-critical tasks (Spark jobs, CI runners)
• Buy Reserved Instances for production baseline capacity, use On-Demand for peaks
• Automatically shut down dev/test environments outside business hours

Storage layer

• Use S3 lifecycle policies: move to IA after 30 days, Glacier after 90 days
• Delete unattached EBS volumes and expired snapshots

Network layer

• Data Transfer Out costs are often a hidden expense
• Keep inter-service communication within the same Region/AZ when possible
• Use VPC Endpoints to avoid NAT Gateway charges

FinOps Practice

Build cost visibility: tag all resources with cost-center and team labels, track cloud spend by team, and set cost alerts.