Separate instants from intervals
Synthetic monitor observes only at 10:00,10:05,10:10. Outage is 10:01–10:04, start included/end excluded. A point detects failure only if it falls inside that interval; there is no independent event stream in this exercise.
| Observation | Inside outage? | Result |
|---|---|---|
| 10:00 | No | Healthy sample |
| 10:05 | No | Healthy sample |
| 10:10 | No | Healthy sample |
Intersect the interruption with observations
- Mark the actual outage interval
- Place each sample instant
- Count observed failures separately
An absent observation cannot be inserted as a healthy sample.
How many failure samples and actual outage minutes are there?
Show the worked answer
Zero failure samples but three actual outage minutes. All three healthy points lie outside 10:01–10:04; they do not prove continuous service.
Bound the observation claim
| Case | Decision | Reason |
|---|---|---|
| Three healthy points | Observed failures 0 | No point falls inside |
| Claim no outage occurred | Unsupported | Unobserved interval remains |
If the operational question concerns short interruptions, supply observations that can distinguish them. Shortening a poll interval changes the model but does not retroactively create evidence.