Service
Infrastructure you are not supposed to think about.
The measure of good infrastructure is how rarely anyone has to look at it. That means unattended updates, deployments that are boring, backups nobody trusts until they have been restored, and alerts that fire before a customer notices.
What this involves
-
Provisioning that is written down, not remembered
A server nobody can rebuild is a liability. Configuration lives in the repository, so the setup is reviewable, repeatable and survives the person who first typed it.
-
Deployment that is boring on purpose
A deployment should be the least interesting event of the day: reproducible, quick to reverse, and without a person following a checklist in a terminal at the wrong hour.
-
Backups measured by restores
A backup that has never been restored is an assumption. What matters is how long a recovery takes and how much data it costs — both known numbers, not hopes.
-
Monitoring and alerting
Uptime, certificates, queue depth, failed jobs, disk. The point is not a wall of dashboards but a small number of alerts that somebody would actually act on.
-
Hardening and keeping current
Key-only SSH access, unattended security updates including third-party package sources, a firewall at one layer rather than two, and no services listening that nothing needs.
Where we would point you elsewhere
We assemble infrastructure that fits the system running on it, which usually means fewer moving parts than people expect. If you want a Kubernetes platform for a team that intends to operate it themselves, hire a platform team — that is a different job, and pretending otherwise would not serve you.
Is this the problem you have?
If it is adjacent rather than exact, say so anyway — the useful conversations usually start that way.