Azure Application Insights for Spring Boot Observability

Application Insights provides distributed tracing, metrics, and log aggregation for Spring Boot services on Azure — the recommended approach is the Java agent, not the SDK, because the agent instruments the application without any code changes.

Java agent — zero-code instrumentation

• Attach the agent as a JVM argument: -javaagent:/path/to/applicationinsights-agent.jar; in App Service, set JAVA_TOOL_OPTIONS=-javaagent:/path/applicationinsights-agent.jar in the application settings
• The agent auto-instruments HTTP requests (request duration, status codes), Spring Data queries (SQL timing, query text), outbound HTTP calls, and exception logging — all without any application code changes
• Application Insights collects this telemetry and correlates it with a distributed trace ID, which means a slow API call can be traced through the Spring Boot service to the database query that caused it

Connection string configuration

• Configure APPLICATIONINSIGHTS_CONNECTION_STRING as an environment variable or application setting — this is the only configuration the agent needs; the connection string includes the instrumentation key and the regional endpoint

Custom telemetry with Micrometer

• The Application Insights Java agent integrates with Micrometer automatically — custom Counter, Timer, and Gauge metrics registered with the MeterRegistry are published to Application Insights as custom metrics
• Use TelemetryClient from the Application Insights SDK for custom events and properties: telemetryClient.trackEvent("OrderPlaced", properties, metrics) adds domain events to the telemetry stream

Querying with Kusto (KQL)

• Application Insights stores telemetry in Log Analytics; queries use Kusto Query Language: requests | where duration > 500 | summarize count() by bin(timestamp, 5m) finds slow requests per 5-minute window
• Create dashboards and alerts directly in Application Insights from KQL queries