In praise of boring infrastructure

There is a particular kind of infrastructure that makes for great conference talks: many moving parts, clever automation, a diagram that needs a second screen. There is another kind that nobody gives talks about, because it works quietly for years. We prefer the second kind.

Complexity is a running cost

Every component in an infrastructure has to be understood, patched, monitored and eventually replaced. A container orchestrator is a fine tool when you operate dozens of services; for a single application with a database, it is mostly a second system to maintain.

What boring looks like

• Every environment can be recreated from code.
• Deployments happen during working hours, because they are routine.
• Backups are restored on a schedule, not just created.
• Alerts fire for things customers notice, and somebody knows what to do.
• A new team member can read how it all fits together in an afternoon.

None of this is novel. All of it is surprisingly rare. And it is the foundation that lets the interesting software on top do its job.