The Device health card shows the current state of a set of devices: the latest value for each metric they report, when it arrived, and whether any of them is alerting. It answers a different question from the rest of the dashboard, which shows what the instruments measured over time.
What it shows
Each device gets a tile carrying its name, a headline reading, and the rest of its metrics beneath with their latest values and timestamps. A metric with no recent value shows a dash rather than a stale number.
A device that is currently alerting is outlined and its breaching value marked, so one tile stands out from the rest. A count at the foot of the card summarises the whole set — how many devices, and how many are alerting.
The search box matters once a card covers more than a handful of devices, since tiles are the widest thing on most dashboards.
Why it belongs on the dashboard
A dashboard of charts can look entirely healthy while telling you nothing, because a chart that stopped updating three days ago looks much like a chart of a stable site. Both are flat lines.
This card closes that gap by showing you the reading and when it arrived. A timestamp from this morning means the instrument is reporting; one from last week means it is not, whatever the chart above it looks like.
Latest value is not the same as within limits
The two things this card shows are independent, and confusing them is the most common way a monitoring programme misreads its own dashboard:
| What you see | What it means |
|---|---|
| Recent timestamps, nothing alerting | Data is arriving and the site is within its limits. |
| Recent timestamps, a device alerting | Data is arriving and something crossed a limit. Act on the measurement. |
| Stale timestamps, nothing alerting | Least obvious and most dangerous. Nothing is alerting because nothing is arriving to evaluate. The site is unmonitored. |
| Stale timestamps, a device alerting | Something crossed a limit and the data is also unreliable. Confirm before acting. |
A current reading says nothing about whether it is within an engineering limit. An alerting device says nothing about whether the instrument is healthy. The third row is the reason this card exists.
Using it well
Put it near the top of any dashboard people open daily, not on a separate page nobody visits. Anyone reading a measurement deserves to know whether that measurement is current, and they will not go looking.
Include the devices that matter to the decision the dashboard supports, rather than every device in the project. A card covering forty devices is a wall of tiles nobody scans.
Pair it with stale data alerts. The card tells you when someone is looking; the alert tells you when nobody is.
Comments
0 comments
Article is closed for comments.