Using Your API ML OpenTelemetry Metrics

Examples of Useability of Telemetry data in API ML

How a system administrator interacts with this data depends on the visualization tool used (e.g., Grafana, Jaeger, or Broadcom WatchTower).

Example 1: High-Level Health Monitoring (Metrics)

A system administrator views a Grafana dashboard. The administrator notices a spike in apiml.request.errors.

• The View: A red line graph shows a sudden jump from 0% to 15% error rate.
• The Insight: By filtering the dashboard using the attribute zos.smf.id, the admin realizes the errors are only occurring on LPAR1, while LPAR2 remains healthy. This suggests a local configuration or connectivity issue on a specific system rather than a global software bug.

Example 2: Latency Troubleshooting (Traces)

A user reports that a specific API is "timing out." The admin finds the relevant traceId in the logs and opens it in a trace viewer.

• The View: A "Gantt chart" style visualization of the request.
• The Insight:
  • apiml.gateway.total: 2005ms
  • apiml.auth.check: 5ms
  • apiml.backend.proxy: 2000ms
• The Action: The admin sees that the Modulith itself only spent 5ms on logic, but waited 2 seconds for the backend mainframe service to respond. The admin can now confidently contact the specific backend service team.