Interactive challenge

LLM Observability and Cost Dashboard Lab

Create a dashboard model that joins user latency, queue wait, GPU pressure, token throughput, and cost signals.

Read matching guide

Difficulty

Medium

Duration

50 min

Persona

SRE

Tools

Prometheus, Grafana, OpenTelemetry

Prerequisites

Metrics and logs stack

Active step 01

Build the signal model

running

Connect user latency to runtime queueing, GPU pressure, token throughput, logs, and traces.

lab@k8sllm:llm-serving

Kubernetes context is loaded. Type commands directly or run the step sequence.