Application insights overview

1. Application insights overview

In this video, we will talk about Application Insights and how it can be used for observability in APIM.

2. Importance of observability

Think of observability like switching on the cockpit instruments in a plane: your API might feel fine from the pilot seat, but Azure Monitor and Application Insights turn dials, gauges, and flight recorders on so you can see truth, not guess. Our goal is to link APIM and your back-ends with a single thread of identity, visualize health on dashboards, and set alarms that wake us before customers do.

3. Telemetry types

The measurements the applications send to Application Insights are known as telemetry. There are three kinds of telemetry: logging, tracing, and metrics.

4. Introduction to logging

Logging is when each event is written in human-readable language. It also comes with a severity level label, which makes it clear whether the event was a normal expected action, an error, or a warning. Imagine a secretary taking notes of a meeting and recording every thing that was said. This is how logging works, but for things that happen inside a software application.

5. Overview of tracing

Multiple related log entries can form a trace. Picture every request as a parcel with a tracking sticker. APIM attaches a unique tracking ID, a correlation ID, into the request headers. Your back-ends, instrumented with the Application Insights SDK, automatically detect this header and include the ID in their own telemetry. This is how the platform can stitch the full journey together.

6. Importance of metrics

Another important telemetry type is metrics. This is an umbrella term that applies to simple measurements, such as counters, average durations, etc. Think of it as a speedometer in your car. It always shows the speed you are traveling at. Metrics can be used for counting errors, average request latency, number of current users logged into the system, etc. They are always measured against time series, so we can see what was happening when.

7. Application insights dashboards

Next comes the command center. Dashboards are your wall of monitors: one glance should answer, “Are we fast, and are we failing?” In Azure Monitor Workbooks or the Azure Dashboard, you'll pin charts for request count, error rate, and latency. Then you turn rules into seatbelts. Create alert rules that watch those charts or the underlying logs, and wire them to an Action Group so the right people get emails, Teams messages, or PagerDuty when thresholds are crossed. Treat SLAs like speed limits: if the p95 latency drifts past your line or the failure rate climbs, the siren goes off before customers start tweeting.

8. Kusto Query Language

When something smells off, the specially designed Kusto Query Language (KQL) is your detective. Open Logs and query the application telemetry like flipping through flight data recorders. With a few saved queries, you can jump from “alert fired” to “root cause” in minutes, then confirm the fix by watching the gauges settle back into the green.

9. Let's practice!

Let's practice what we learned!

Create Your Free Account

By continuing, you accept our Terms of Use, our Privacy Policy and that your data is stored in the USA.