Why Boring Infrastructure Wins in Production

Every system I run in production is built on boring tech. DynamoDB. SES. Lambda. Route 53. Next.js. React. Six months ago I might have been tempted to reach for something cooler. Now I actively resist. The boring stack is the one I can still debug at midnight, and that is the only quality that matters.

What "Boring" Means

Boring is not a style preference. Boring is a set of properties:

Well-documented: I can find an answer to any question in five minutes
Battle-tested: thousands of engineers have hit the same bugs
Stable API: the thing I learned two years ago still works today
Predictable: performance and cost do not surprise me
Hireable: if I need help, I can find someone who knows it

By that definition, Postgres is boring. DynamoDB is boring. React is boring. Those are compliments.

What Is Not Boring

The latest hyped-up database that promises 10x performance. The framework that got 40k stars last month. The orchestration system that only the creator's company actually runs in production. Those are all exciting. They are also expensive to adopt in ways that are not obvious on day one.

The cost of non-boring tech is not the learning curve. It is the long tail: every weird bug is a new adventure, every hire has to be trained, every year the ecosystem pivots and you pivot with it.

The One Exception

I make exactly one exception. If a new tool is a 10x improvement on the critical path of my business, I will adopt it and pay the cost. AI code assistants were that kind of exception 18 months ago. Almost nothing else has been since.

My Current Stack

plaintext

Backend:   FastAPI on Lambda (5+ years old)
Database:  DynamoDB (14+ years old)
Frontend:  Next.js on Vercel (8+ years old)
Email:     SES (14+ years old)
DNS:       Route 53 (15+ years old)
CI/CD:     GitHub Actions (6+ years old)

Every piece on that list is multiple years old, stable, well-documented, and unlikely to change substantially in the next five years. That predictability is the entire point. I can spend my thinking on the product, not the plumbing.

The Discipline

The hard part is resisting the urge to rewrite when something shinier shows up. I keep a "tempting tools" list in my notes. Every entry has to survive six months before I will consider adopting it. Most do not. The ones that do have a real place in the stack. The boring stack wins because it never stopped working.

The Choose Boring Technology essay by Dan McKinley is required reading.