Meraki DataMagic - Wireless Health Overview

Overview

The Wireless Health Overview is a new, health-focused view (currently in beta) of how your Meraki wireless is performing. It answers a simple question: is this network healthy, and if not, where is the problem? It does this without you having to work through the detailed estate tables.

The overview works at every level of your estate: start from the whole estate or an organisation for a summary and a ranked list of networks, or focus on a single network or access point for the full detail. You choose the scope, and the time range, at the top of the page.

Choosing what to look at

• Scope - click the selector at the top left to open the estate tree, then pick the whole estate, an organisation, a network or an individual access point. An up-arrow next to the selector steps you back up a level, for example from a network back to its organisation, without reopening the tree.
• Time range - the pills at the top right (6h, Day, Week, Month and 2mo) set the window used for the health score, funnel, trends and signal figures. A wider range gives a smoother, longer-term picture; six hours is the shortest, chosen so it always includes the most recent readings.

Starting from the whole estate or an organisation

When you select the whole estate or an organisation, rather than a single network, the overview shows a rollup: a table of the networks in that scope, ranked worst-health-first so the ones needing attention are at the top. For each network you see:

• Health - the network's overall Good, Fair or Poor rating.
• Connection - the percentage of clients that connected.
• Performance - its combined performance rating.
• RF - Meraki's RF score for the network.
• Clients - the number of wireless clients connected.
• Alerts - the number of open Meraki health alerts.

A network's Connection figure shows a dash rather than a percentage when fewer than five clients were seen trying to connect in the window, because a reading from only a handful of clients is too thin to be meaningful. The network is still rated on Health, Performance and RF from the data that is present. Select any network in the list to open its own view (described below); the up-arrow brings you back to the rollup. Above the table, a summary shows how many networks are Good, Fair and Poor, a filter box lets you find one by name, and a pager steps through the list 25 at a time.

At organisation level you also get the overall health score, category cards, connection funnel and trends described below, rolled up across all of that organisation's networks. The whole-estate view instead lists networks from every organisation together, with those counts in place of a single score, since combining separate organisations into one figure would not be meaningful.

Health score and category ratings

The large ring on the left is the overall health score, from 0 to 100, weighted across the four categories shown beside it. Each category is rated Good, Fair or Poor (green, amber or red) and shows the figures behind the rating. Hover any card to see what it measures and how it is rated.

• RF Health - the quality of the radio environment (Meraki's RF score).
• Connection - the percentage of clients that connected. This counts clients, not connection attempts, so a single device retrying in a loop counts once and can no longer drag the figure down while everyone else is connecting fine.
• Performance - latency, signal-to-noise, packet loss and channel utilisation combined.
• Network Service - how well the services behind every connection (authentication, DHCP and DNS) are working, for the clients that got past association. The card's heading names the service that left the most clients stuck, and the figure beneath it is the share of clients that service stranded, so here a lower figure is better. The card's colour reflects the wider picture, the share of clients that get through once past association, so green means almost everyone is connecting and red means many are not.

Where a category has no data (for example, some networks do not report an RF score, or fewer than five clients were seen trying to connect in the window), the card shows a dash, and the overall score is worked out from the categories that are present.

Connection funnel

The connection funnel shows how wireless connection attempts progress through each step (association, authentication, DHCP and DNS) and how many reach success. Each bar shows the share of attempts still succeeding at that point and, when the connection is not healthy, the step that loses the most attempts is highlighted, so you can see at a glance where connections drop off. Hover any step for what it means, and click it to see the failures logged there.

Attempts, not clients. The funnel counts connection attempts, whereas the Connection card above counts clients, so the two figures can legitimately differ. One misconfigured device retrying in a loop makes thousands of failed attempts - which shows up in the funnel - yet it is still only a single client, so it barely moves the Connection card. Read the Connection card for how many of your clients actually got on, and the funnel for where in the sequence attempts are dropping off.

Trends

The Trends section plots key wireless metrics over the selected time range as small line charts, so you can tell whether things are improving or getting worse. Hover a trend to see its latest value and range.

Access points

The Access points list shows every wireless access point in the network, sorted worst-health-first so the ones needing attention are at the top. It has the same summary counts, filter and pager as the networks list. For each access point you see:

• Clients - the number of wireless clients connected.
• Ch. util - the busiest radio's channel utilisation.
• Loss - the worse of its downstream and upstream packet loss.
• SNR - its signal-to-noise ratio (higher is better).
• Health - an overall Good, Fair or Poor rating for the access point.
• Status - whether it is online.

Drilling into the detail

Two parts of the overview open a slide-in panel with more detail:

• Click an access point in the list to see its downstream and upstream packet loss, per-radio channel utilisation and latency, its signal-to-noise ratio, and its own trends over time.
• Click a step in the connection funnel to see the failures logged at that step: the access points and clients with the most failures, and how the failures vary over time.

How the ratings are worked out

Every reading is compared against a threshold to decide whether it counts as Good, Fair or Poor. The thresholds follow Meraki's own guidance for healthy wireless:

Metric	Good	Fair	Poor
Signal-to-noise ratio	27 dB or higher	20 to 27 dB	below 20 dB
Latency	up to 60 ms	60 to 150 ms	above 150 ms
Packet loss	up to 1%	1 to 3%	above 3%
Channel utilisation	up to 20%	20 to 50%	above 50%
Clients connected (Connection and Network Service)	90% or higher	50 to 90%	below 50%

Meraki's RF score, the Performance rating and the overall gauge are rated on the same 0-to-100 scale as 80 or higher Good, 50 to 80 Fair, and below 50 Poor. The client-based Connection and Network Service ratings are held to a stricter bar - 90 or higher Good, 50 to 90 Fair, below 50 Poor - because "nine in ten clients connected" is the sensible line for a healthy wireless network. For Connection this bar is applied to the percentage the card shows. For Network Service it is applied to the share of clients that connect once past association, which sets the card's colour, rather than to the smaller "clients stuck" figure the card prints. A category's rating combines the metrics within it, and the overall score is a weighted blend of the four categories, so a single poor reading pulls down both its category and the overall score, which is exactly what you want when you are hunting for problems.

Where the figures come from

The overview draws on more than one Meraki source, which is worth knowing when two numbers do not line up exactly:

• The health score, category cards, funnel and trends come from Meraki's Wireless Health measurements, collected roughly every two hours and rolled up over the time range you have selected. The Connection and Network Service cards use the per-client version of that measurement, which is why they are counted in clients.
• In the access points list, Clients, Ch. util and Loss are each access point's live last-hour readings from the device records, while SNR is averaged over the selected time range.
• In the networks ranking, Health, Connection, Performance and RF are rolled up over the selected time range, while Clients and Alerts are the latest values - so those two columns can move independently of the rating.
• The connection funnel and its drill-down are attempt-based and, for the drill-down, come from Meraki's separate failed-connections log. Because that is a different feed, its totals can differ from the funnel's; a network can show a dip in the funnel but few individual log records, or the reverse.