When you integrate Istio with User Workload Monitoring for metrics, you need to create a PodMonitor resource in each of your dataplane namespaces in order for your workloads to send metrics to User Workload Monitoring. It would be nice if this were automated for you in some way, for example, by having the PodMonitor resource created automatically in namespaces where a sidecar is running.

Using the Cluster Observability Operator instead of UWM should also be investigated.

Another aspect we should investigate is how we could improve usability around tracing integration.