Monitoring for Microservices

The consumption needs for microservices are not transparent to customers. They do not know what to set the resources to in the microservice manifest, leading to over provisioning and confusion. Real time understanding of the CPU/RAM usage of each microservices would provide the information needed.

It is possible to set up microservice monitoring as the Cloud Ops team has done this via Prometheus/Grafana. Stefan Wischtel has written about this on TechCommunity. However, this requires additional development efforts from the customer's developers. Ideally, this information should be available within Cumulocity IoT.

In addition, this will reduce resource consumption within Cumulocity IoT. Currently, we have customer microservices that are using 0.005 CPUs, while the manifest file states they can use 1 whole CPU. This leads to a lot of unnecessary resources. By providing this information to the customer, we will be able to better optimize our platform and reduce our costs.

Use cases:

• As a business user, I want to understand the resource requirements for my microservices, so that I can minimize potential invoices.

• As a technical user, I want to monitor consumption for my microservices so that I can analyze trends and find anomalies.