Call Drops

Call Drops is a fleet-wide anomaly report. It surfaces the NAPs dropping more calls than usual, compared against each NAP's own history.

Open from TB Nodes > Call Drops.

Call Drops — top-N tables per drop type and window

---

Layout

The page is a grid of top-N tables — one table per drop type × time window:

Drop Type	What it means
Local	The NAP dropped the call locally.
Remote	The remote peer dropped it.
System	Dropped by the unit itself (resource exhaustion, routing failure, etc.).

Time Window	Purpose
15m	Spot something happening right now.
1h	Catch issues that started in the last hour.
24h	Daily rollup.
7d	Weekly pattern.

Each table lists the top NAPs by drop count over that window, with how unusual the volume is compared to normal.

---

Baseline & Anomaly Detection

Each NAP is compared against its own 12-week time-of-week baseline. "Monday at 2pm" is compared to previous Mondays at 2pm — so an evening maintenance window doesn't look like an anomaly just because traffic was light.

A z-score flags statistically elevated drops:

• High z-score = dropping much more than this NAP normally does at this hour.

• Low or zero z-score = within normal range for the NAP's usual pattern.

Confidence Flags

• Limited confidence — the baseline has too few samples (new NAP, recently added host). The value is shown but trust it less.

• Insufficient data — not enough history to compute a z-score at all. The score is omitted.

---

Drill-Through

Click a NAP name in any table to open its NAP Detail page — all 27 metrics for that NAP on that device, plus a configurable chart to investigate the drop pattern over time.

NAP Detail drill-through view