If you've ever sat in a technical leadership or architectural role at a cloud-native company, you've likely witnessed the classic clash between Engineering and the Finance department. On one side, developers fighting to refactor bottlenecks. On the other, a well-intentioned "FinOps" team trying to slash the AWS bill armed with nothing but Excel spreadsheets.
The outcome of this story is usually predictable (and frustrating): real code optimization projects get shelved because the finance team chose to sign a 3-year contract for Reserved Instances (RIs) or Savings Plans to "solve the cost issue."
The Illusion of Financial Discounts
Let's be pragmatic: financial optimizations and RIs are excellent management instruments. The problem arises when we use long-term contracts as a band-aid for inefficient architectures.
Buying reserved capacity on AWS to run a service suffering from chronic memory leaks, or that executes thousands of N+1 database queries, is literally subsidizing waste at a discount. You aren't solving your system's scale problem; you're merely ensuring the company pays slightly less for continuous waste over the next three years.
Cost-Aware Architecture
As a Senior Architect, I argue that true FinOps isn't just done in the AWS billing portal; it's done in the IDE, in domain design, and in infrastructure choices.
Cloud costs are no longer a mere consequence of the system; they have become a non-functional requirement, just as critical as latency, security, or high availability. In a Cost-Aware Architecture, the financial impact of every microservice or asynchronous worker is designed and validated from Day 1.
This is exactly why, strategically, I advocate for the introduction of languages like Rust in critical processing domains. The low memory footprint and absurdly efficient CPU usage of Rust code drastically change the baseline of the required infrastructure.
The Right Value Stream
Architectural efficiency must always precede financial negotiation. Only after eliminating the computational bottleneck — whether by redesigning inter-service communication or injecting a high-performance language into the data plane — does it make sense to bring in the finance team to sign Reserved Instance contracts based on the newly optimized metrics. The ROI of optimizing the code first is incomparable.
In short, reducing the cloud bill isn't a war between Engineering and Finance. It's a partnership where Engineering tackles computational waste, and Finance tackles commercial waste. If your company is only doing the latter, you are leaving a lot of money on the table.
Top comments (0)