A tiny reliability lab · Synthetic data
Data pipelines need manners.
A small, browser-only simulation of three useful ingestion habits: validate before accepting, make retries safe, and explain rejected events. No live service or real operational data is involved.
Choose an event
Accepted: 0
Rejected: 0
Last event: Waiting for a reading.
| Event | Reading | Result | Reason |
|---|---|---|---|
| No events yet. Try the valid reading. | |||
What this shows
Validation keeps malformed values out of trusted records. A stable event ID lets the receiver recognize a retry instead of counting it twice. A clear rejection reason gives operators something they can act on.
This is a teaching example, not a production pipeline benchmark. Real systems also need durable queues or storage, authentication, monitoring, replay policies, and domain-specific validation.
Read the operational observability case study ↗