......................................................................................... 3 1. Metrics Comparison...................................................................................... Benchmarking........................................................................................ 3 1.2 Metrics Details......................................................................................... ..........................................................................18 3 I. Overview 1.Metrics Comparison 1.1 Features Benchmarking Features KubeSphere OpenShift Rancher Monitoring ※※※※※ ※※※※

17h 10.42.0.6 rke2-s1 kube-system helm-install-rke2-metrics-server-nvcp6 0/1 Completed 0 17h 10.42.0.3 rke2-s1 11h 10.42.1.0 rke2-a1 23 kube-system rke2-metrics-server-78b84fff48-jvnnd 1/1 Running 0 12h 10.42.2.0 17h 10.42.0.6 rke2-s1 kube-system helm-install-rke2-metrics-server-nvcp6 0/1 Completed 0 17h 10.42.0.3 rke2-s1

cluster. Kubernetes monitors the clusters at multiple levels. Heapster is used to aggregate vital metrics, while the kubelet node agent queries cAdvisor to fetch data from containers and provide to Heapster the core component that scrapes and stores data. • Prometheus node exporter: gets host level metrics and exposes them to Prometheus. ©Rancher Labs 2017. All rights Reserved. 43 DEPLOYING DEPLOYING AND SCALING KUBERNETES WITH RANCHER Host Stats The host stats dashboard shows the metrics related to the host, as gathered by Prometheus. Prometheus Stats The Prometheus

Users can access Grafana and see metrics for the resources to which they have access. They can also annotate their workloads to have Prometheus begin to scrape custom metrics from them. 3.3.4.2 OpenShift the same stack. In a multi-cluster level, RHACM offers a single pane of glass view of cluster metrics of OpenShift clusters using popular open source projects like Grafana, Observatorium and Thanos 4 Anthos Anthos enables application observability through Anthos service mesh. Cluster-level metrics have limited support through Cloud Logging and Cloud Monitoring or Prometheus and Grafana. Both

ingress # ingress: provider: nginx kubernetes_version: v1.14.9-rancher1-1 monitoring: provider: metrics-server # # If you are using calico on AWS # # network: # plugin: calico # calico_network_provider: ingress # ingress: provider: nginx kubernetes_version: v1.15.6-rancher1-2 monitoring: provider: metrics-server # # If you are using calico on AWS # # network: # plugin: calico # calico_network_provider: ingress # ingress: provider: nginx kubernetes_version: v1.16.3-rancher1-1 monitoring: provider: metrics-server 33 # # If you are using calico on AWS # # network: # plugin: calico # calico_network_provider:

weave_node: "" weave_cni: "" pod_infra_container: "" ingress: "" ingress_backend: "" metrics_server: "" windows_pod_infra_container: "" ssh_key_path: "" ssh_cert_path: "" ssh_agent_auth:

weave_node: "" weave_cni: "" pod_infra_container: "" ingress: "" ingress_backend: "" metrics_server: "" windows_pod_infra_container: "" ssh_key_path: "" ssh_cert_path: "" ssh_agent_auth:

elasticsearch_exporter • …… 注：相比于基于Exporter的这种通用解决方案，业界一些成熟的开发框架或平台也会提供更加原生的监控解决方案，比如针 对Spring社区Spring Boot开发框架Metrics采集的micrometer-registry-prometheus组件。此外，还有一些调用链解决方 案也能做到更加全面细粒度的监控，比如目前比较流行的Skywalking、Pinpoint等，类似的还有一些成熟的商业解决方案。