The thing I learned last week that helped me really understand was – you can split linux tracing systems into

• data sources (where the tracing data comes from)
• mechanisms for collecting data for those sources (like “ftrace”)
• tracing frontends (the tool you actually interact with to collect/analyse data)