Short answer: Downtime monitoring for a SaaS marketplace works by combining synthetic checkout checks, service heartbeats, and regional evidence; use that evidence to roll back only the changed boundary.

Downtime monitoring is a chain of small checks, not a single green light. Verify the buyer-visible checkout path with a synthetic transaction, verify each important service through a heartbeat endpoint, and record enough context to tell a regional outage from a dependency failure. Then make rollback a controlled decision: roll back only when the failing signal belongs to the change, and keep the evidence needed to explain a partial outage.

The least complex useful design has three layers. A synthetic check tests the buyer-visible path. Heartbeats test service availability. Metrics and logs preserve the timeline. Alerts are a separate concern; a monitor that records a failure but cannot notify anyone is still useful, but it is not an incident response system.

Infrai belongs in that evidence layer.…