Cloud Infrastructure Cost Optimization
Cloud bills grow faster than business — this is normal without active management. Typical situation: company pays for dev environments running 24/7, oversized instances, unused EBS volumes, old snapshots. Audit usually gives 20-40% savings without losing performance.
Where costs hide
Compute (EC2/GCE/VM). Biggest line item. Problems:
- Oversized instances (c5.2xlarge for service under 200 RPS)
- Dev/staging running nights and weekends
- Old instances on previous generations (r4 instead of r6i)
Storage. Silently accumulates:
- EBS volumes from deleted instances (orphaned volumes)
- Snapshots older than 90 days (often stored for years)
- S3 objects without lifecycle policy
- Suboptimal storage class (Standard instead of Infrequent Access for archives)
Data transfer. Data transfer out is expensive. Especially: cross-AZ traffic, egress from region.
Idle resources. Load balancers without traffic, NAT Gateways, Elastic IPs without attachment.
Analysis tools
AWS Cost Explorer: built-in, free. Shows spending by services, regions, tags.
AWS Trusted Advisor / GCP Recommender: specific recommendations: "This instance uses < 5% CPU — consider t3.small instead of t3.xlarge".
Infracost: analyze Terraform plan costs before applying. In CI/CD shows cost diff for each PR.
Timeline
Audit and analysis — 2-3 days Quick wins — 2-3 days Rightsizing plan + implementation — 3-5 days Reserved/Savings Plans purchase — 1 day