Reducing Cloud Costs with FinOps: A CTO's Practical Guide · Daniel Tinizaray

There are two types of CTOs: those who look at their AWS bill and know exactly what every dollar corresponds to, and those who hold their breath and pay.

If you're in the second group, you're not alone. According to the FinOps Foundation 2025, 67% of organizations report that their cloud costs are growing faster than revenue. Worse: 40% of cloud spend is wasted on unused or poorly sized resources.

I've spent years optimizing cloud infrastructure on AWS and OCI for companies in Ecuador and the US. This article shares the practical FinOps framework I apply with every client — no theory, just what works.

What Is FinOps (and Why Should You Care)?

FinOps isn't "cost optimization." It's a cultural and operational discipline where engineering, finance, and product collaborate to maximize the value of cloud spend. It's not about turning off servers — it's about making sure every dollar spent on cloud has a measurable return in delivery velocity, availability, or scalability.

The 5 Cloud Cost Reduction Levers

After auditing dozens of AWS and OCI accounts, these are the levers that consistently deliver the biggest impact:

1. Strategic Tagging + Showback

Without tags, your cloud bill is a black box. Implementing mandatory tagging by team, project, and environment is step zero of any FinOps initiative.

Required tags:
  - cost-center: [marketing, engineering, data]
  - environment: [production, staging, dev]
  - project: [checkout-redesign, data-pipeline]
  - owner: [team-name]
  - auto-shutdown: [true/false]

With tags in place, you implement showback: each team sees their consumption and cost. Simple visibility reduces spend 10-20% without touching a single instance.

2. Reserved Instances & Savings Plans

Stable workloads (databases, application servers) should never run on-demand. A 3-year RDS reserved instance on AWS costs ~60% less than its on-demand equivalent.

• Compute Savings Plans: 66% savings vs. on-demand. Flexibility across instance families.
• EC2 Instance Savings Plans: 72% savings for predictable workloads.
• OCI Universal Credits: 30-50% discount vs. PAYG with portability across regions.

3. Spot Instances + Spot Fleets

For interruption-tolerant workloads (batch processing, CI/CD, staging, workers), Spot Instances cost up to 90% less than on-demand.

On AWS, use EC2 Spot Fleets with a mix of instance types to minimize interruptions. If one AZ runs out of spot capacity, the fleet auto-failovers to another available instance type.

4. Continuous Right-Sizing

Over-provisioning is the most common sin. Most EC2 instances run at 10-30% capacity. An m5.xlarge (4 vCPU, 16 GB) costs $140/month when a t3.medium (2 vCPU, 4 GB) at $35/month would handle the load.

Tools: AWS Compute Optimizer, OCI Cloud Advisor. My recommendation: quarterly right-sizing for production, monthly for dev/staging.

5. Auto-Shutdown + Schedules

Dev and staging environments typically run 24/7 when they're only used ~50 hours a week. A simple shutdown schedule (7 PM → 7 AM, weekends off) reduces these environment costs by 60-70%.

# Example with AWS Instance Scheduler
# Dev/Staging: 07:00-19:00, Monday-Friday
# Estimated savings: from $1,200/mo to $360/mo

FinOps on OCI (Oracle Cloud Infrastructure)

OCI has specific features that work in your favor if you know them:

• Universal Credits: Pre-pay model with 30-50% discounts vs. PAYG. Plus: unused compute credits can be used for storage or networking.
• OCI Cost Analysis: Native tags with cost tracking per compartment. Superior to many third-party tools.
• Bring Your Own License (BYOL): If you're coming from on-premise Oracle Database, migrating to OCI saves significant licensing costs.

FinOps Culture: The Missing Ingredient

I've seen companies with all the tools still overpay because no one owns the cloud cost. FinOps is a culture as much as a practice. Rules I implement:

• Every feature includes a cloud cost estimate in the engineering ticket
• Budget alerts: 80%, 90%, 100% + per-service over-usage
• Weekly cost dashboard by team vs. budget
• Shared responsibility: devs don't just deploy — they also optimize

TL;DR — What You Can Implement This Week

1. Audit your current bill — Identify services >$100/month and ask "is this generating value?"
2. Implement mandatory tagging — No visibility, no optimization
3. Shut down dev/staging at night — 60-70% immediate savings
4. Evaluate Savings Plans — For stable workloads, 60%+ discount
5. Measure, measure, measure — What gets measured gets optimized

Does your cloud bill feel like a black hole? Book a call and we'll audit your bill in 30 minutes. No cost, no obligation, no financial jargon.