Metoro has two main components: the in-cluster agents and the observability backend. At a very high level, the flow of data in Metoro is as follows:
  1. The node agents are responsible for collecting data from the linux kernel of all nodes in the Kubernetes cluster and writing to cluster local storage.
  2. The cluster exporter then reads the data from the local storage aggregating across all nodes and sends it to the Metoro backend ingesters along with the Kubernetes metadata (pods, deployments, configmaps, etc).
  3. The ingesters write all observability data to the long-term backend storage: currently clickhouse.
  4. The api server reads data from the backend storage and serves it to the frontend and any api clients.
The following diagram shows the high level architecture of Metoro: