Threshold Configuration

Thresholds define the quality boundaries that determine a probe's health status. After each probe execution, every configured metric is compared against its thresholds. If any metric breaches a threshold, the probe's health status changes and configured webhooks fire. This guide covers how thresholds work, which metrics support them, and best practices for setting values.

How Thresholds Work

Each metric has two threshold levels:

After each probe execution, the platform evaluates every metric against its thresholds:

1. Retrieve the latest metric values from the completed probe call
2. Compare each metric against its warning and critical thresholds
3. Determine the individual result for each metric (Healthy, Degraded, or Unhealthy)
4. The worst result across all metrics determines the overall probe health status

This means a single metric breaching its critical threshold makes the entire probe UNHEALTHY, even if all other metrics are within normal range.

Understanding Metric Direction

Metrics fall into two categories based on whether higher or lower values indicate better quality:

Higher Is Better (Inverted Metrics)

For these metrics, quality decreases as the value goes down:

Example: MOS warning = 3.5, MOS critical = 2.5

• MOS 4.1 = HEALTHY (above 3.5)
• MOS 3.2 = DEGRADED (below 3.5, above 2.5)
• MOS 2.1 = UNHEALTHY (below 2.5)

Lower Is Better (Standard Metrics)

For these metrics, quality decreases as the value goes up:

Example: Jitter warning = 40ms, Jitter critical = 50ms

• Jitter 25ms = HEALTHY (below 40ms)
• Jitter 45ms = DEGRADED (above 40ms, below 50ms)
• Jitter 55ms = UNHEALTHY (above 50ms)

Configurable Metrics and Defaults

The following metrics support threshold configuration:

The default values provide a reasonable starting point based on industry standards and ITU-T recommendations. You should tune them based on your specific quality requirements.

Adjusting Thresholds

1. Open your probe's detail page
2. Navigate to the Settings tab
3. Locate the Thresholds section
4. Modify warning and critical values for each metric
5. Click Save

Health Status Determination

The overall probe health status follows this logic after each execution:

For each metric with a configured threshold:
  1. Retrieve the measured value from the probe call
  2. Compare against warning and critical thresholds
  3. Assign individual metric status:
     - Within acceptable range --> HEALTHY
     - Exceeds warning but not critical --> DEGRADED
     - Exceeds critical --> UNHEALTHY

Overall probe status = WORST individual metric status

If no data was collected: status = UNKNOWN

Examples:

Status Transitions and Webhooks

When the probe's health status changes from one state to another, configured webhooks are triggered:

See Webhooks for details on webhook payload format and delivery.

Consecutive Failure Counting

To avoid alerting on transient network glitches, you can configure consecutive failure counting. Instead of changing health status on a single threshold breach, the probe requires N consecutive breaches before transitioning.

Example with consecutive count of 3:

This prevents false alerts from single-execution anomalies while still catching persistent problems.

Best Practices for Setting Thresholds

Start with Industry Standards

The ITU-T G.107 recommendation and industry experience provide these guidelines:

Tune Based on Your Environment

After running probes for a week, review the metric history:

• If metrics consistently hover near the warning threshold, either your infrastructure needs attention or your threshold is too tight
• If metrics never come close to warning, you can tighten thresholds for earlier detection
• Set warning thresholds at the point where quality becomes noticeable to users
• Set critical thresholds at the point where calls become unusable

Account for Geography

Probes running from distant regions will naturally have higher RTT. Adjust RTT thresholds accordingly:

Example Configurations

Enterprise voice (strict quality requirements):

Standard SIP trunk monitoring:

Best-effort / development environment:

Next Steps

• Webhooks -- Get alerted when thresholds are breached
• Creating a Probe -- Set up a new probe with these thresholds
• Status Pages -- Share probe health status publicly
• Probe Health States -- Complete health state reference