1. Providing observability for your components means exposing information about their inner state so that you can access the data easily and reason about the actual state of your components. In other words, if something is observable, you can understand it.
2. WMI Exporter can be used to monitor a Windows node host OS and hardware. For monitoring the Docker Engine itself, you can use the experimental metrics server exposed by the Engine.
3. In production environments running at a large scale, you can use Prometheus Operator to easily deploy and manage multiple Prometheus clusters for different needs.
4. WMI Exporter and the Docker Engine metrics server are exposing the metrics on dedicated ports on each node. We need two extra scraping jobs that handle them individually.
5. Use the Telegraf service hosted directly in your container...