Service and workload entities are separating in Entity Catalog and Entity Graph
Service and workload entities are separating in Entity Catalog and Entity Graph
Application services and the Kubernetes workloads they run on are now separate entities, connected by a relationship, instead of a single merged entity. This follows the model most teams already work with: an application runs on infrastructure, and in Kubernetes that infrastructure is a workload.
Getting your stack mapped is more flexible with different instrumentation setups. Previously, a service entity was assembled by matching labels across your application and infrastructure telemetry. A service with imperfect instrumentation could be missing from the graph entirely, even when you could see it in Application Observability. Each entity now appears on the strength of its own telemetry.
Kubernetes workloads that were previously discovered as services are hidden by default. A setting in the Entity Catalog includes them when you want to see workloads alongside your services.
Insights are unchanged. Insights on a workload still roll up to the service running on it, so your existing troubleshooting and alerting workflows carry over.
This applies to newly created Grafana Cloud stacks.
