azure-monitoring

Azure Monitoring & OpenTelemetry

OpenTelemetry Setup (.NET)

builder.Services.AddOpenTelemetry()
    .ConfigureResource(r => r.AddService("MyApp"))
    .WithTracing(t => t
        .AddAspNetCoreInstrumentation()
        .AddHttpClientInstrumentation()
        .AddEntityFrameworkCoreInstrumentation()
        .AddOtlpExporter())
    .WithMetrics(m => m
        .AddAspNetCoreInstrumentation()
        .AddHttpClientInstrumentation()
        .AddOtlpExporter());

Application Insights

// Connection String (nicht Instrumentation Key!)
builder.Services.AddApplicationInsightsTelemetry(options =>
{
    options.ConnectionString = builder.Configuration["APPLICATIONINSIGHTS_CONNECTION_STRING"];
});

Umgebungsvariablen

Variable	Lokal (Aspire)	Produktion (Azure)
OTEL_EXPORTER_OTLP_ENDPOINT	http://localhost:4317	—
APPLICATIONINSIGHTS_CONNECTION_STRING	—	InstrumentationKey=...
OTEL_SERVICE_NAME	Auto (Aspire)	Manuell setzen

Custom Metrics

private static readonly Meter AppMeter = new("MyApp.Business");
private static readonly Counter<long> OrdersCreated = AppMeter.CreateCounter<long>("orders.created");
private static readonly Histogram<double> OrderProcessingTime = AppMeter.CreateHistogram<double>("orders.processing_time_ms");

Alert-Design

• SLO-basiert: „99.9% der Requests unter 500ms" statt „CPU > 80%"
• Severity korrekt: Sev0 = Ausfall, Sev1 = Degradation, Sev2 = Warning
• Action Groups: Teams-Channel für Sev0/1, Email für Sev2
• Alert-Fatigue vermeiden: Aggregation, Suppression Windows