Spotting LAGGED clusters on the timeline chart

The timeline chart in the LAGGED Summary Report is useful precisely because it turns a table of individual Agent lag events into a visual pattern. A list of 200 rows tells you that lag happened; the chart tells you whether those 200 events arrived in one 20-minute window or trickled in over two weeks. That distinction drives completely different responses.

What the chart axes represent

The horizontal axis is time, bucketed to match the date range you selected. A single-day range typically shows hourly bars. A multi-day range shows daily bars. The vertical axis is the count of LAGGED events in each bucket — every event where the server logged that an Agent session took longer than expected to acknowledge a status update.

The shape of the chart matters as much as the total count. A tall bar surrounded by low bars is a one-off event. Bars that rise together across two or three consecutive time buckets are a cluster — and clusters are what you are hunting.

Three cluster patterns and what they mean

• Morning or afternoon cluster repeating each day — your peak dial window is overloading the server. The campaign Dial level or the number of Agent seats logged in at the same time is too high for the box's current CPU and memory allocation. This is the most common pattern and the most actionable: you either lower the dial level or resize.
• A one-day spike in an otherwise flat chart — something changed on that specific day: a new campaign launched, a backup ran during dial hours, or a database job competed with Asterisk for I/O. Look at what changed operationally on that date.
• A rising baseline that climbs week over week — gradual capacity erosion. As your call volume grows and your server hardware stays constant, the average lag count drifts up. The chart makes this visible weeks before agents start complaining.

Narrowing from cluster to cause

Once you have located a cluster on the chart, note the start and end time of the raised bars. Use that window as your filter in the Agent LAGGED Report. There you will see the server IP, user ID, campaign, and call Status (lead status) for each event in that window. If one server IP dominates, the issue is hardware-side. If events spread evenly across all server IPs but cluster on a specific campaign, that campaign's settings are the cause.

flowchart TD
  A["Cluster identified on timeline chart"] --> B{"Repeats same time each day?"}
  B -->|Yes| C["Peak dial period overloading server"]
  C --> D["Lower dial level or resize box"]
  B -->|No| E{"Isolated to one date?"}
  E -->|Yes| F["One-off event — check change log for that day"]
  E -->|No| G{"Rising week over week?"}
  G -->|Yes| H["Gradual capacity erosion — plan upgrade"]
  G -->|No| I["Drill into Agent LAGGED Report for that window"]

Using the chart proactively, not reactively

Most operators look at this chart after agents complain. A better use is a weekly five-minute check before complaints arrive. Run a rolling two-week view every Monday. If the highest bar this week is taller than the highest bar last week, investigate now. The chart gives you a week or two of lead time before lag becomes a call-quality problem that agents can feel.

Cross-referencing with server load

A LAGGED cluster on the timeline and a server load spike at the same hour are strong confirmation that the box, not the network, is the limit. If the server load graph is flat during the same window, the lag may trace back to a Latency issue on the agent's endpoint rather than the server. Checking both charts together takes two minutes and eliminates half the possible root causes.

For a broader routine covering all the metrics worth checking each week, see our server health and capacity guide. For detail on the server load side of the comparison, VICIdial system load explained covers exactly what that number means.

If your timeline chart shows a cluster every single afternoon, that is a sizing problem, not a configuration problem. See what a VICIfast managed box costs — our plans are sized to match common agent seat counts, and every box is live in under 40 seconds.