data-api-builder-observability

name	data-api-builder-observability
description	Wire Data API Builder telemetry to Application Insights, Azure Log Analytics, or any OpenTelemetry backend.
license	MIT

Data API Builder Observability

Use when

• Tracing REST, GraphQL, MCP, and database activity end-to-end.
• Shipping metrics (request count, errors, duration, active requests) to Azure Monitor or a collector.
• Diagnosing latency or error spikes across scaled-out DAB instances.

Workflow

1. Stand up a destination: Application Insights, an OTLP collector/backend, or Azure Log Analytics with DCE/DCR/custom table.
2. Configure runtime.telemetry: application-insights, open-telemetry, azure-log-analytics, and optional log-level.
3. Use dab add-telemetry for App Insights/OTel; use dab configure for Log Analytics and log-level filters.
4. Start DAB, drive traffic, and confirm requests, dependencies, traces, metrics, or custom-table logs arrive.

Minimal config

"runtime": {
  "telemetry": {
    "application-insights": { "enabled": true, "connection-string": "@env('app-insights-connection-string')" },
    "open-telemetry": { "enabled": true, "endpoint": "http://otel-collector:4317", "headers": "@env('otel-headers')", "exporter-protocol": "grpc", "service-name": "dab" },
    "azure-log-analytics": {
      "enabled": true, "dab-identifier": "my-dab-instance", "flush-interval-seconds": 10,
      "auth": { "custom-table-name": "DabLogs_CL", "dcr-immutable-id": "dcr-...", "dce-endpoint": "https://my-dce.eastus-1.ingest.monitor.azure.com" }
    },
    "log-level": { "default": "Warning", "Azure.DataApiBuilder.Core": "Information" }
  }
}

Guardrails

• Allow a graceful shutdown window; ephemeral containers can exit before OTel exports flush.
• Don't log connection strings or PII via custom headers; treat headers as a secret.
• Application Insights uses connection strings; avoid bare instrumentation keys except for legacy compatibility.
• OpenTelemetry options are configured via dab add-telemetry, not dab configure.
• Log Analytics requires Managed Identity with Monitoring Metrics Publisher on the DCR and a custom table ending _CL.
• App Insights Learn KQL uses requests, dependencies, traces, and customDimensions; custom Log Analytics tables use suffixed columns like Identifier_s.

Related skills

• data-api-builder-config
• data-api-builder-health
• data-api-builder-caching
• aspire-data-api-builder

Microsoft Learn

• https://learn.microsoft.com/azure/data-api-builder/concept/monitor/open-telemetry
• https://learn.microsoft.com/azure/data-api-builder/concept/monitor/application-insights
• https://learn.microsoft.com/azure/data-api-builder/concept/monitor/log-analytics
• https://learn.microsoft.com/azure/data-api-builder/concept/monitor/log-levels