The application emits the business-relevant metrics via HTTP. The Prometheus instance scrapes and store this data and makes it available via queries, graphs, and external solutions, such as Grafana.

In a container orchestration, the Prometheus instance runs inside the cluster. This removes the necessity to configure externally accessible monitoring endpoints. Prometheus integrates with Kubernetes in order to discover the application instances. Prometheus needs to access every application pod individually, since every application instance emits its monitoring metrics separately. Prometheus accumulates the information of all instances.

The Prometheus configuration is either stored in a config map or part of a base image. The instance is configured to access the applications and exporters every n seconds in order to scrape the time series. For configuring Prometheus, refer to its current documentation.

This is one possible solution for integrating business monitoring into a cloud native application.

Business-related metrics are advisably represented by domain events that emerge as part of the business use case. Integrating the chosen monitoring solutions should happen transparently from the domain logic, without much vendor lock-in.