This is the module that will change how you think about production. Not incrementally, fundamentally. Before this module, production is a black box that either works or does not. After it, production is a system you can interrogate, understand, and reason about in real time.
Logs, metrics, and traces are not three competing choices. They are a maturity curve. Logs tell you what happened. Metrics tell you how often and how fast. Traces tell you why -- following a single request through your load balancer, into Rails, through Active Record, into Postgres, back up through the middleware stack, and out to the client. Each layer of observability answers questions the previous layer cannot.
Brendan Gregg's Systems Performance book is dense. It is 800 pages of how operating systems, CPUs, disks, and networks actually behave under load. It is also the most referenced book in production debugging conversations among serious infrastructure engineers. When the curriculum says READ, it means read. You will not absorb it all on the first pass. You will return to specific chapters every time you face a new class of performance problem, for years.
OpenTelemetry is the future of observability instrumentation, and the future is arriving now. It is a vendor-neutral standard for generating traces, metrics, and logs. Learn it before your company adopts it and you are scrambling to instrument your services in a sprint. Rails has emerging OpenTelemetry support, and the Ruby SDK is production-ready. Being the engineer who can set up distributed tracing across services is a staff-level differentiator.
The Grafana + Prometheus + Loki stack is free, powerful, and what most teams that are not paying for Datadog end up running. Build dashboards for your own applications. The act of deciding what to measure teaches you more about your system than any code review.