A cloud security dashboard shows a firewall protecting data between a computer and database, with network diagrams and monitoring charts.
Azure Network Watcher's Connection Monitor is getting a troubleshooting upgrade. The feature is in public preview, and it puts the health and metrics of individual Azure hops in the same place as the failing test. The Azure Updates listing carries the title "[In preview] Public Preview: Azure Network Watcher – Resource health and topology in connection monitor tests" and a 17 November 2026 timestamp. When we looked, the live Azure Updates page for that ID showed no records, so those details come from the submitted metadata. The behavior described below comes from Microsoft Learn's Connection Monitor documentation.

A cloud security dashboard shows a firewall protecting data between a computer and database, with network diagrams and monitoring charts. What Connection Monitor already does​

Connection Monitor is Network Watcher's continuous connectivity tester. It lets you monitor connectivity and latency between endpoints inside and outside Azure. Tests run over HTTP, TCP or ICMP. Each test reports two numbers:

  • the percentage of checks that failed
  • the round-trip time (RTT)

The topology view draws the network path between source and destination. Until now, a test could tell you that loss or latency had changed. It was less direct at showing which Azure component along the path was responsible.

What the preview adds​

Microsoft's documentation says you can now select an Azure resource hop in the test topology and see its health metrics in the same view. The idea is to line up the test's packet loss and latency against each hop's metrics and find the hop behind the drop in connectivity.

The documented workflow is:

  1. Open the test page and go to the network topology. Each Azure resource hop shows its Resource health status below its name and IP address.
  2. Select an Azure resource hop in the path.
  3. A pane opens on the right with the resource name and a Metrics section. It shows the resource's metric charts for the selected time range.
  4. Compare those charts with the test's checks-failed percentage and RTT over the same period. Select Performance charts to see the test's trend lines.
  5. Select the other hops one at a time.
  6. If you need more detail, go to the resource from the topology, or use its Network Insights view for detailed metrics and alerts.

The metrics in the pane are the same resource metrics that Azure Monitor Network Insights shows for that resource type. Microsoft's example is Azure Firewall, where the pane can show Data processed, Throughput and Firewall latency ping. The documentation points to an "Availability of resources" list of supported resource types. That list wasn't in the material we reviewed, so check it before assuming your hop type is covered.

Limits to know before relying on it​

  • It's a preview. Microsoft warns that features and functionality might change before general availability. Azure's own definition of "In preview" is availability to all customers for non-production use and testing. Treat it as a diagnostic aid, not something to build runbooks around yet.
  • Azure hops only. Resource metrics aren't shown for on-premises or internet hops. A missing pane on those hops does not mean they are healthy.
  • Some hops have no latency value. When the source or destination is inside Azure, the topology merges two things: a traceroute and an internal logical-route lookup. That lookup is similar to Network Watcher's next hop diagnostics. Traceroute usually can't identify hops inside Azure boundaries, and the logical route isn't a latency measurement. As a result, a few merged hops, mostly inside Azure, show no latency. That is a quirk of how the topology is built, not proof of zero latency.
  • It's a correlation tool. The documentation describes matching a test's loss and latency to hop metrics. It doesn't promise automatic root-cause detection. You still have to read the test and hop charts together.
  • Other details weren't confirmed. The material we reviewed doesn't mention extra setup steps, pricing, or changes to quotas for this preview, so don't assume any.

Why it matters for admins​

Picture a hybrid app where a web tier in Azure reaches a backend through a firewall and a gateway. A Connection Monitor test starts failing 8% of its checks, and RTT climbs. Before, you would open the topology, spot a suspect hop, then hop into separate resource blades or Network Insights to check metrics. Now you can click the firewall hop and compare its throughput and latency charts with the test's trend over the same window. If they move together, you have a lead. If they don't, you move to the next hop.

Microsoft's FAQ also says a test can breach its loss or RTT threshold with no issues found on any hop. This feature won't change that. It gives you more evidence to check, not a guarantee of an answer.

Background that helps when reading results​

  • Test states: Pass, Fail, Warning, Indeterminate and Not Running. With no thresholds set, Connection Monitor assigns defaults of 750 ms RTT for TCP or ICMP tests and 10% checks failed. Warning triggers when the checks-failed percentage exceeds 80% of a threshold you've set. With no threshold specified, Warning triggers once the automatic one is exceeded. A test with no threshold reaches Fail only at 100% failed checks.
  • Data storage: Connection Monitor stores data in the Log Analytics workspace you pick at creation. It also surfaces metrics in Azure Monitor, which keeps them for 30 days by default. For longer historical comparisons, rely on Log Analytics retention.
  • Scale limits: 100 connection monitors per subscription per region, 20 test groups per monitor, 100 sources and destinations per monitor, and 20 test configurations per monitor.
  • Sources and agents: Sources need the Network Watcher extension (Azure VMs and scale sets) or Azure Arc-enabled hosts. Destinations such as URLs, FQDNs and IPs don't need the extension.

Practical takeaways​

  • Try the preview on a non-critical monitor first and confirm that your resource types appear in the supported list.
  • Always compare charts over identical time windows. Mismatched windows produce misleading correlations.
  • Don't read a blank pane on an internet or on-premises hop as a clean bill of health.
  • Keep the Log Analytics workspace and alert rules as your production signal. Use the new topology pane for quick triage.

Overall, this is a modest but useful change. It shortens the path from "something is slow" to "this Azure hop looks off", as long as you remember it only covers Azure hops and is still a preview.

 

References

  1. (In preview) Public Preview: Azure Network Watcher – Resource health and topology in connection monitor tests Azure Updates 2026-11-17T20:00:32Z
  2. Connection Monitor Overview - Azure Network Watcher | Microsoft Learn learn.microsoft.com
  3. Network Watcher Frequently Asked Questions (FAQ) learn.microsoft.com