The evolution of monitoring at Nutanix: Combining the best of a legacy system with Kubernetes, Prometheus, and Grafana
As Nutanix has begun moving from running services in virtualized environments to running containerized applications in microservices environments, its monitoring systems have evolved too. Its new Kubernetes-based Microservices Platform has a monitoring stack leveraging Prometheus Operator, Prometheus, Alertmanager, and Grafana. Currently, Nutanix has applications running in both the VM-based and microservices environments, and Rohith Subramanyam, Senior Member of Technical Staff, will talk about how Nutanix has bridged the two.
Over the years, Nutanix has built an in-house monitoring system called Prism Alertmanager, which displays alerts to customers on Prism, the management console. It has a number of advanced features that the Prometheus Alertmanager used in the Microservices Platform does not. In order to provide a single pane of glass where customers can view and manage both the alerts being generated by traditional services as well as Microservices Platform, the team built a system using the operator pattern that can forward Prometheus alerts to the Prism alerting system. Subramanyam will walk through this non-trivial work that the team took on to come up with a solution.