Latency Monitor Guide: Cards, Sparklines, Trends & Sorting

Quick answer: open Monitor to watch the same live measurements as the map, laid out as cards. Each card is one destination with its latest millisecond value, a colour and a sparkline. Above the cards you get a summary (Up, Down, Average, Slowest right now), a strip of things that need attention and a trend chart of your overall average. Use it when you care about numbers and history more than geography.

Why a second view of the same data?

A map is excellent for "where". A card grid is excellent for "how much" and "since when". Because both use one engine and one selection, you can switch between them with no setup. Your history is kept, so the sparklines are already populated when you switch.

The summary row

Item	Meaning
Up	Destinations whose latest reading succeeded
Down	Destinations whose latest reading failed or timed out
Average	Mean of the latest readings across the destinations that replied
Slowest right now	The destination with the highest latest reading, with its value

Needs a look

This strip lists only destinations that are down or in a warning state. When everything is normal it stays quiet. It is the quickest answer to "is anything wrong?" on a wall display or a second screen.

Anatomy of a card

• Name and category of the destination.
• Latest value in milliseconds, coloured by state.
• State text such as Normal, Slower than usual, Much slower than usual or No reply.
• Sparkline: the last readings drawn as a tiny line, so you can see trend and spikes at a glance.
• Average, best and worst of the stored readings (up to 60).
• No-reply %: the share of readings that failed. A destination at 0% is perfectly reachable; 5% or more points to intermittent loss or blocking. See packet loss.

Clicking a card opens the same detail panel as the map pin, including the 95th percentile and a personal alert limit for that destination.

How states are decided

State is relative to that destination's own recent average. Once at least three earlier readings exist, a reading more than 40% above the average (default) is yellow, more than 70% above is red, and a failure is down. A rise smaller than 30 ms is never flagged. You can change both percentages in settings, from 10 to 100% for yellow and 20 to 200% for red. A per-destination limit overrides the global one. Details are in Alerts and thresholds.

The trend chart

The trend chart plots the average latency across all selected destinations over time. A flat line means a steady connection. A step up that persists means something changed, like a network switch. A sudden spike that returns to normal is a transient event. Because it averages across many servers, it is the best view for spotting local problems: if the average jumps while only a few cards change, the cause is likely far away; if every card changes together, the cause is near you.

Sorting, filtering and compact view

• Sort: problems first (default), slowest, fastest, name, or the original list order.
• Filter: type to narrow the list by name or category.
• Compact view: smaller cards so many destinations fit on one screen, ideal for wall displays.

Controls

You can pause and resume the engine, trigger a check immediately, and open settings for interval and timeout. While a Test is running in the other mode, the live engine pauses automatically and resumes afterwards, so a test is never disturbed by Monitor traffic.

Worked example: finding the cause of evening lag

You pick 15 destinations and leave Monitor open from 17:00. Until 19:00 the trend chart is flat near 45 ms and every card is green. At 19:20 the chart climbs to 120 ms and stays there until 22:00; all 15 cards turn yellow or red at the same time, and the no-reply rate rises from 0% to 6%. Because all destinations changed together, the problem sits on your side of the internet, typically your Wi-Fi, a busy household upload or evening congestion at your ISP. If only two Asian destinations had changed, the problem would be a route far away. With that evidence, see troubleshoot high latency.

Tips for good monitoring

• Choose 10 to 30 destinations that represent what you actually use: your games, your cloud region, your CDN.
• Leave the interval at 30 seconds unless you are hunting short spikes.
• Include at least a few nearby destinations as a reference: if nearby ones are slow, the problem is local.
• Use the 95th percentile in the detail panel rather than the worst value when judging spikes. See percentiles.

Common mistakes and myths

• Treating Monitor as an uptime service. It runs only while your tab is open; it is not a 24/7 server-side monitor.
• Expecting history after clearing browser data. It lives in local storage.
• Reading the average card as the world average. In Monitor, "average" is your own recent average.

What to do next

Set up alerts in Alerts and thresholds, compare with the live map, or open Monitor now.