How do we do Ops with containers?

Although this is not a book for system administrators or site reliability engineers, you should know the basic context in which containers are generally run. The main idea is that containers are largely stateless “functions” that process inputs (web requests or HTTP messages from other services) and produce outputs (web responses, side effects, and logs streamed to STDOUT). In a well-run operations environment, containers can be thought of as an analog to Linux processes, or to functions in programming.

Containers are usually “scheduled” onto hosts by a third-party tooling layer such as Kubernetes, Nomad, and others. If containers are like processes, then these fill the role of the operating system scheduler (the whole thing is a distributed system instead of a single host).

Container output is usually captured by the same tooling and redirected to logging solutions such as Logstash, Graylog, and Datadog. Metrics...