Databases Status
Everything is up.
Databases and data platforms usually sit directly on the request path, so even seconds of unavailability turn into immediately visible errors in the applications in front of them. That immediacy is also what makes them easy to blame unfairly: the symptoms of a genuine outage and the symptoms of an exhausted connection pool are almost identical from the application's point of view.
Managed platforms add two wrinkles worth knowing. Control plane and data plane are separate, so branching, provisioning, or dashboard access can break while production queries keep being served perfectly. And on serverless or scale-to-zero tiers, a cold start is normal behaviour that looks exactly like a hang, which means a first slow query after an idle period is usually not an incident at all.
Warehouses and streaming platforms fail more quietly. Buffering absorbs short disruptions, so the visible symptom is lag rather than failure, and the damage only shows up once dashboards or downstream systems have fallen behind.
- up
- slow
- down
- no reading
Common questions about databases and data platforms
Check your connection count first, because exhausting a pool produces errors nearly indistinguishable from an outage and is by far the more common cause. Then try connecting from a different network or client. If a fresh client from elsewhere succeeds, the platform is fine and the problem is your pool, your credentials, or an IP allowlist.
On serverless and scale-to-zero tiers the compute suspends when idle and has to resume, which can take seconds. That is documented behaviour rather than an incident. If every query stays slow after the first one, that is worth investigating.
Usually yes, but a limited one. Managed platforms run their control plane separately from the data plane, so console access, branching, and provisioning can be unavailable while queries continue to be served normally. It blocks operational work rather than your users.