In Praise of Boring Infrastructure

I’ve come to treat “boring” as the highest praise I can give a piece of infrastructure. Not outdated, not lazy — boring: predictable, well-understood, unsurprising, the kind of thing that does its job so quietly you forget it’s there. It’s an unfashionable thing to admire in a field that rewards the new and the clever, but I’ve spent enough nights being surprised by exciting systems to believe the goal of good infrastructure is to be as boring as possible, and to spend your novelty budget very carefully.

Novelty is a cost, and you pay it in incidents

Every new, clever, or cutting-edge piece you put into a load-bearing position is a cost, even when it’s genuinely better on paper. It’s a thing fewer people understand, with fewer battle-worn answers when it breaks, with failure modes nobody’s mapped yet, and with a smaller pile of “someone already hit this exact problem” waiting for you at 3am. Boring, mature technology has had its sharp edges found and filed down by everyone who came before you. New technology hands you the job of discovering them — in production, usually, at the worst possible time.

Every clever choice is a loan against future incidents. Sometimes it’s worth taking. But it’s a loan, and boring infrastructure is how you keep the debt low where it matters most.

This isn’t an argument against ever using new things. It’s an argument for knowing you’re spending when you do, and spending it where the payoff is real rather than where you were just bored.

Spend your novelty budget where it wins, not everywhere

Think of it as a budget: you get a limited amount of “interesting” before the cognitive and operational cost catches up with you. So spend it on the part of the system that’s actually your differentiator — the thing that’s the whole point — and be aggressively boring everywhere else. Use the well-trodden database, the boring language, the standard deployment approach for the plumbing, so you can afford to be innovative in the one place innovation earns its keep. Teams that spend novelty everywhere end up with a stack where every layer is a little surprising, and the surprises compound into a system nobody fully understands. Boring plumbing is what buys you the right to be interesting where it counts.

Boring is a property you build, not one you’re stuck with

Here’s the part that reframes it: boring isn’t the absence of effort, it’s the result of it. A system becomes boring because someone made it predictable — because the failure modes were rehearsed instead of extrapolated, because the upgrade path is quorum-aware, because the monitoring is treated as production too, because the runbook was written and tested. The calm is earned. An “exciting” system is often just one where that work hasn’t been done yet, so it still has the capacity to surprise you. The work of operations is, in large part, the steady conversion of exciting systems into boring ones.

The homelab taught me to want the boring version

I learned to value this by running things at home where they’re allowed to be exciting — a homelab is where you meet the messy parts and collect the gremlins on purpose. That’s exactly the right place for novelty: low stakes, nobody paged, failure is the curriculum. It’s also what makes you crave the boring version for anything that matters, because you’ve felt how much attention “interesting” actually demands. The lab is where you spend novelty freely; the load-bearing system is where you hoard it.

Boring scales; clever fights you

The quiet superpower of boring infrastructure is that it stays out of your way as things grow. Predictable systems compose predictably. Well-understood pieces have well-understood interactions. You can add load, add people, add scope, and the boring foundation just keeps being boring. Clever systems, by contrast, tend to get more surprising under growth — the edge cases multiply, the one person who understood it becomes a bottleneck, and the cleverness that saved a little up front charges compound interest later.

So when someone describes a system I built as boring, I take the win. It means it does its job, it doesn’t wake anyone up, and its novelty budget went where it mattered instead of getting sprinkled everywhere for the thrill. Exciting infrastructure is a phase; boring infrastructure is a destination, and reaching it is most of the work. If you’ve deliberately made something more boring and slept better for it, I’d like to hear about it.