Skip to main content

Maintenance Services

An overview of the categories of service that make up this platform's maintenance layer, independent of which specific tool fills each one. Each category is compared in detail in the Services catalog; which tool actually runs here, and how it was set up, is in Walkthrough → Services → Maintenance.

Deployment platforms​

A deployment platform sits on top of the server and handles the operational layer — running services, managing SSL, routing traffic, and providing a UI so not everything requires the command line. CI/CD sits right next to it — building and testing code before it ever reaches the platform — a distinct concern, but close enough that the two sometimes blur into one tool.

Development​

Tools for code hosting, review, and collaboration.

Security​

This is where Context → Security stops being an abstract cross-cutting concern and becomes a concrete category of tool. Three related but distinct roles. Password manager is where a person's own credentials live, built for a human to unlock. Credential manager is where infrastructure's own secrets live, built for a machine to read at runtime. Auth / IDP is a centralised identity provider handling login across all services — one account, one place to manage access. The same tool rarely does all three well, which is why they stay separate concerns even though they're all solving the same underlying problem: keeping a secret out of the wrong hands.

Monitoring​

Three layers, narrow to broad. Stats cover CPU, memory, disk, and container status — some of this comes free from the deployment platform itself; a dedicated tool goes deeper and keeps working even if that platform is what's having problems. Error tracking / APM covers what's actually going wrong inside an application, not just whether the box is up. Status pages are the public-facing view — what's up, what isn't, communicated to whoever depends on it.

Backups​

This is where Processes → Maintenance → Backups stops being reasoning and becomes a concrete category of tool, same as Security does for its own Context page. Two modes, not one job done twice: disaster recovery is off-site, slow, and covers everything at once; operational recovery is local, fast, and covers the common case of just needing to roll back one instance.