Monitoring (System Insight)

Metrics from all layers – From Kubernetes (providing metrics from services and pods), and from Usage Engine (providing metrics from EC, Platform, and workflows) – are easy to send to a Unified Monitoring Layer like Prometheus. Metrics from Usage Engine are of two types: 

  • MIMs (simple metrics): these are gauge or sum metrics with one fixed label only. Only metric name and value can be customized. There is a set of automatically generated MIMs, but a user may also add customized MIMs. MIMs have fixed publishing intervals. 
  • Prometheus UDRs (complex metrics): for use cases where several labels – that can vary – are needed for a metric, Prometheus UDRs are used. Any parameters in the UDR, including time stamp, can be customized. 

Metric Monitoring Approach

MIMs and Prometheus UDRs are sent to a Servlet, where they are scaped regularly by Prometheus.  

For visualization, Grafana can be used. Usage Engine provides a set of predefined Dashboards for System Statistics. Also, in the Operations UI, a base set of System Statistics for the past 10 minutes can be viewed.