Gap Duration — VoIP Quality Metric
RTCP-XR average gap period duration in ms — short gaps between SIP bursts mean insufficient recovery time, compounding audio degradation across the call.
Gap Duration
| Property | Value |
|---|---|
| Key | gap_duration |
| Unit | ms |
| Type | Gauge |
| Direction | Receive |
| RFC | RFC 3611 Section 4.1 (RTP Control Protocol Extended Reports) |
What It Measures
Gap Duration is the average duration of gap periods — the stretches of time between burst loss events when the network is delivering packets normally. A gap duration of 2,000ms means that on average, each clean period between bursts lasted about two seconds before the next burst began.
Gap duration is the RTCP-XR counterpart to Burst Duration. Together they describe the rhythm of loss on a call: how bad each burst event is and how much recovery time the endpoint gets between them.
Why It Matters
Gap duration determines whether the listener can recover psychologically and perceptually between loss events. Short gap duration is a signal that bursts are frequent and closely spaced — the functional equivalent of continuous degradation even if individual bursts are brief.
Consider the difference:
- Long gap duration (above 3,000ms): Bursts are infrequent. Between each event, the listener and the jitter buffer both have time to stabilize. Codec concealment works well. Most users accept this level of quality.
- Short gap duration (below 500ms): Bursts are closely spaced. The jitter buffer cannot drain before the next burst arrives. Concealment events stack up. Users experience what feels like continuous quality degradation even though the burst density metric looks moderate.
For SIP infrastructure testing at scale, gap duration often drops before burst density rises. It's an early signal that your network is struggling — bursts are becoming more frequent even before they become longer or more severe. Use it as a capacity headroom indicator during load tests.
Thresholds
| Level | Value | Meaning |
|---|---|---|
| Good | Above 3,000ms | Long recovery periods, infrequent bursts |
| Warning | 500ms – 3,000ms | Frequent bursts, jitter buffer under pressure |
| Critical | Below 500ms | Persistent degradation, near-continuous loss events |
How to Fix It
- Read alongside Burst Duration. Short gap duration combined with long burst duration is the worst-case scenario — frequent, lengthy loss events. Address burst sources before tuning anything else.
- Check for periodic congestion. Gap duration that clusters around a fixed interval (e.g., every 2 seconds) suggests a periodic background process competing for bandwidth — a backup job, a sync operation, or a monitoring agent that polls heavily.
- Increase jitter buffer depth. If gap duration is short, the jitter buffer needs to be larger to provide a buffer against rapidly succeeding bursts. This trades latency for stability.
- Improve path diversity. Short gap duration with long burst duration often indicates a single-path failure mode. Adding route diversity or failover paths can reduce burst frequency and lengthen gaps.
- Validate codec FEC settings. For Opus streams, enabling FEC provides in-band redundancy that can span short gaps without requiring jitter buffer intervention.
Related Metrics
- Burst Duration — Average length of each burst event
- Burst Density — Fraction of call time spent in burst loss state
- Gap Density — Fraction of call time in the good (gap) state
- Jitter — High jitter forces smaller gap duration by compressing effective good periods