I’ve been building a Kubernetes operator called SLOK to manage Service Level Objectives using CRDs and Prometheus.
The goal is to make SLOs a first-class, declarative resource in the cluster. You define objectives, targets and windows, reference a PromQL-based SLI, and the controller periodically evaluates it, calculates the error budget, and updates status accordingly. The current implementation focuses on percentage-based SLIs and simple validation, with more advanced features planned (threshold SLIs, burn rates, alerting, templates).
This is still an early version and very much a learning project, but I’m curious how this approach resonates with others who’ve implemented or operated SLO systems on Kubernetes.
Feedback on the overall design, scope, or comparison with existing tools would be appreciated.
Show HN: SLOK – A Kubernetes operator for declarative SLOs and error budgets | Heykuki News