Performance analysis with the dashboard
My favorite tool by far when I just want to know what's going on in the cluster is the Kubernetes dashboard. There are a couple of reasons for this, as follows:
It is built-in (always in sync and tested with Kubernetes)
It's fast
It provides an intuitive drill-down interface from the cluster level all the way down to individual container
It doesn't require any customization or configuration
While Heapster, InfluxDB, and Grafana are better for customized and heavy-duty views and queries, the Kubernetes dashboard's pre-defined views can probably answer all your questions 80–90% of the time.
You can also deploy applications and create any Kubernetes resource using the dashboard by uploading the proper YAML or JSON file, but I will not cover this because it is an anti-pattern for manageable infrastructure. It may be useful when playing around with a test cluster, but for actually modifying the state of the cluster, I prefer the commandline. Your mileage may...
