Smart Quorum Consensus: Fewer False Alerts
Multiple probes alone are not enough. You still need a clear rule for turning their results into one trustworthy verdict. That is why nsmon does not declare an outage because of a single isolated failure. It confirms incidents only when a majority of the contacted probes agree within the same decision window.
What Smart Quorum Consensus actually does
Every check in nsmon is built from parallel measurements taken from multiple locations. Smart Quorum Consensus defines how those separate results become one final status. This dramatically reduces false alerts caused by local noise, a single provider problem, or a limited regional network issue.
From a public product perspective, the principle is simple: if the service fails only from one location, we do not automatically treat that as confirmed service downtime. We confirm an incident only when a majority of the contacted probes agree within the same time window.
When quorum logic protects you from false positives
One probe has a local problem
Without quorum logic, one failed result could trigger an unnecessary alert. With quorum, that result is judged against the others.
The issue exists only in one region
If the service works from most other locations, we do not want to classify that as confirmed service downtime.
The network has a brief burst of noise
A short anomaly on part of the probe set does not always mean a real incident affecting the service as a whole.
The service is genuinely unavailable across locations
When most contacted probes see the same failure, that becomes a strong signal of a real outage.
How the quorum decision works in practice
The service is checked from multiple probes
nsmon contacts several geographically separated probes. How many are involved depends on the subscription tier.
Results are compared in the same decision window
What matters is how the service behaves across the contacted locations in the same monitoring moment, not one isolated result on its own.
A majority is required for a confirmed outage
If a majority of the contacted probes agrees that the service is unavailable, that becomes a confirmed incident.
An isolated regional problem should not wake you up
If the service still works from most locations and the issue is limited to one narrow part of the internet, we do not want to present that as a false global outage of your service.
How to think about the quorum verdict
In practical terms, the key question is how many probes see the service as healthy and how many see it as unavailable at the same time.
Contacted probes 5 Probes reporting OK 1 Probes reporting FAIL 4 Result Incident confirmed
The probe count depends on the plan
A higher subscription tier means more parallel viewpoints and stronger confidence in the verdict.
One isolated failure is not enough
A single probe failure should not be able to declare a full incident on its own.
The goal is confidence, not hypersensitivity
Monitoring should be fast, but it should not drown users in false positives.
A regional issue is not automatically your outage
If the problem appears only in one network or one part of the world, we do not want to treat that as a confirmed global outage of your service.
Why we do not alert on an isolated issue in one network
If the service works from most locations and the problem affects only one provider or one region, that often is not something the service owner can directly control.
That is why we do not want to create panic every time one network path degrades. The goal is to alert on situations that really represent service unavailability for a broader portion of users.
Single-check monitoring vs quorum-based decisions
| Topic | Simple single-check approach | Quorum approach in nsmon |
|---|---|---|
| Decision input | One result from one location. | Multiple results from several probes in the same time window. |
| Behavior under noise | Easy to trigger false alerts. | Much better at filtering local spikes and isolated failures. |
| Handling uncertainty | Often too sensitive and too absolute. | Better at reducing unnecessary alerts when the evidence is weak or limited. |
Quorum only works when you have multiple independent probes.
The global probe network guide explains why nsmon collects results from multiple locations and why one probe alone is not enough.
Read the global probe network guide βTypical quorum situations
One location reports failure while the others are healthy
That usually is not enough to confirm an incident. Preventing those unnecessary alerts is exactly what quorum logic is for.
Most contacted probes report the same failure
That is a strong signal that the problem is real and not just a local network issue.
The service works from most regions but not from one
That can still be serious for some users, but it does not automatically mean a confirmed global outage of the service as a whole.
Multiple probes agree during the same decision window
That is how you get a trustworthy incident confirmation instead of a random one-off failure.
Important limitations
- β Quorum logic cannot replace well-designed monitoring or a well-chosen endpoint.
- β Some regional problems are real for part of the audience, even if they do not qualify as a confirmed global outage in public monitoring.
- β The depth of confirmation depends on how many probes are used, which depends on the subscription tier.
- β The goal of quorum logic is to reduce false positives, not to model every isolated internet issue perfectly.
What to focus on when evaluating the result
Multiple probes across different regions The agreement between independent viewpoints matters more than one isolated result.
Compare results between locations This helps separate a local anomaly from confirmed service unavailability.
Parallel results from the contacted probes A confirmed incident should be based on simultaneous evidence from multiple locations, not on one late or isolated failure.
Frequently asked questions
Alerts only matter when they are based on a trustworthy decision.
nsmon combines results from multiple probes and confirms incidents with quorum logic, so you get fewer false alerts and a clearer signal when something is really wrong. Create a free account and monitor with more confidence.