The cloud brought unprecedented flexibility to companies, allowing resources to be provisioned with clicks. However, without active governance, this ease translates into inflated cloud bills and operational waste that erode business margins. This is where FinOps (Financial Operations) comes in.
In this article, we will cover practical techniques that our engineering team applies in our cloud infrastructure and FinOps consultancy to audit and optimize costs in AWS environments.
1. Identifying Oversized Instances (Right-sizing)
The most common mistake in cloud infrastructures is overprovisioning. Developers often choose instances with more CPU and RAM than the application actually needs "just in case."
How to act:
- Monitor historical CPU and memory usage using tools like AWS Compute Optimizer or CloudWatch.
- If an EC2 instance's average CPU utilization is below 20% for more than 14 days, downgrade to a smaller family (e.g., from
t3.xlargetot3.medium). - Consider migrating to Graviton-based instances (g families), which offer better cost-performance ratios for Node.js and databases.
2. Cleaning Orphaned Resources and Idle Storage
Many companies pay for resources that are no longer connected to any active system.
What to audit:
- Orphaned EBS Volumes: When an EC2 instance is deleted, the storage disks (EBS) often continue to exist and be charged. Delete volumes that are in the
availablestate. - Unused Elastic IPs (EIP): AWS charges for static public IPs that are reserved in your account but not associated with an active instance.
- Old Snapshots: Poorly configured backup policies generate daily snapshots that are never deleted, quietly accumulating costs in S3.
3. Strategic Use of Reserved Instances and Savings Plans
For workloads that run 24/7 (such as production servers), paying the on-demand price (On-Demand) is the most expensive option.
The strategy:
- Apply Savings Plans or purchase Reserved Instances to get discounts of up to 72% in exchange for a commitment to use them for 1 or 3 years.
- Leave the on-demand price only for temporary testing or staging environments, which can be turned off outside business hours.
Conclusion
FinOps is not a one-time project, but a culture of continuous improvement. By regularly auditing these pillars, your business keeps the infrastructure fast and resilient, ensuring that every cent invested in the cloud generates real value for the operation.