Grafana Alloy📜
Overview📜
Grafana Alloy, formerly known as Grafana Agent, is Grafana’s opinionated spin of the OpenTelemetry collector. It combines many open-source projects in the cloud-native observability space with the goal of being the only observability component necessary to collect and distribute telemetry signals within a cluster.
flowchart TD
subgraph Monitoring
Prometheus/Thanos
Loki
end
subgraph DT[Distributed Tracing]
Tempo
end
subgraph A[Alloy]
Alloy
Alloy ==> |Traces| Tempo
Alloy ==> |Metrics| Prometheus/Thanos
Alloy ==> |Logs| Loki
end
style EU stroke-dasharray: 10 10
subgraph EU[End-User Applications]
App-A -->|OpenTelemetry| Alloy
App-B -->|OpenTelemetry| Alloy
App-C -->|OpenTelemetry| Alloy
end
subgraph N[K8s Node]
CL[Container Logs]-->|Logs|Alloy
NE[Node Exporter]-->|Metrics|Alloy
end
subgraph ServiceMonitors
Service-A<-->|Metrics|Alloy
Service-B<-->|Metrics|Alloy
Service-C<-->|Metrics|Alloy
end
Big Bang Touchpoints📜
Licensing📜
Grafana Alloy is open-source, licensed under Apache 2.0.
UI📜
While Grafana Alloy does expose a UI for visualizing its configuration status, it is not necessary for use and is not exposed by default within Big Bang.
Storage📜
Grafana Alloy requires no storage itself, opting instead to push telemetry signals to other cluster components like Loki and Tempo, which have their own storage needs.
Logging📜
Grafana Alloy writes its logs to stderr. These logs will be picked up by the logging collector configured within the cluster.
High Availability📜
Grafana Alloy supports multiple deployment modes with built-in clustering.
Depending on which features are enabled in the k8s-monitoring
chart, Alloy
may be deployed as a StatefulSet
, DaemonSet
, or Deployment
.
Health Checks📜
Grafana Alloy is configured with standard liveness and readiness probes. In addition to the health of Alloy itself, cluster administrators can view the UI mentioned above for specific health statuses of individual Alloy components.