In the preceding chapters you learned about all the components in a Prometheus setup: instrumentation, dashboards, service discovery, exporters, PromQL, alerts, and the Alertmanager. In this final chapter you will learn how to bring all of these together and plan a Prometheus deployment and maintain it in the future.
When you are considering a new technology it’s best to start the rollout with something small that doesn’t take too much effort, nor prematurely commit you to doing a complete rollout. When starting with Prometheus in an existing system, I recommend starting by running the Node exporter1 and Prometheus. You already ran both of these in Chapter 2.
The Node exporter covers all the machine-level metrics that might be used from other monitoring systems, and then quite a few more as was covered in Chapter 7. At this stage you will have a wide variety of metrics for little effort, and you should get comfortable with Prometheus, set up some dashboards, and maybe even do some alerting.
Next, I’d suggest looking at what third-party systems you are using and which exporters exist for them and start deploying those. For example, if you have network devices you can run the SNMP exporter, if you have JVM-based applications such as Kafka or Cassandra you would use the JMX exporter, and if you want blackbox monitoring you might use the Blackbox exporter, as covered in Chapter 10. The goal at this stage is to gain metrics about ...