Drift detection and autoscaling share the same observe-compare-act loop

A developer building a deployment platform realized that drift detection and autoscaling are structurally identical, both following an observe-compare-decide-act loop. The key insight is that actual system state must always be read from the provider, not from internal database records, since a database row marked 'running' can persist even after the underlying VM has been deleted. Missing baseline data is another common pitfall — comparing against a non-existent desired state produces false positives that erode team trust in alerts. The author also notes that storing state in unqueried metadata columns defeats the purpose of recording it at all. Treating both features as instances of the same contract enables shared scheduling logic and a consistent, infrastructure-free testing strategy.