References: Structured Logging for LLM Calls
-
Logging (computing) - Wikipedia - Foundational coverage of logging including structured-vs-unstructured tradeoffs and the JSON-line pattern used in this chapter's canonical schema.
-
JSON - Wikipedia - Reference for the JSON serialization format that every structured-logging schema in this textbook uses for log lines.
-
OpenTelemetry - Wikipedia - Coverage of the vendor-neutral observability standard that defines the LLM semantic conventions referenced in this chapter and Chapter 10.
-
Observability Engineering - Charity Majors, Liz Fong-Jones, and George Miranda - O'Reilly - The canonical reference for high-cardinality structured logging; Chapter 4 on event-driven debugging maps directly to this textbook's per-request log line approach.
-
Site Reliability Engineering - Beyer, Jones, Petoff, Murphy (eds) - Google / O'Reilly - The free Google SRE book has chapters on monitoring distributed systems that inform the schema design covered here.
-
OpenTelemetry GenAI Semantic Conventions - OpenTelemetry - The official semantic conventions for LLM telemetry including the gen_ai.* attribute names that should appear in any vendor-neutral logging schema.
-
The Twelve-Factor App: Logs - Adam Wiggins - The Heroku-era manifesto on treating logs as event streams; concise statement of the log-as-data-source mindset this chapter assumes.
-
Honeycomb Blog: Structured Events - Honeycomb - Working notes on high-cardinality structured logging from the team behind one of the leading observability platforms; the structured-events posts pair well with this chapter.
-
Datadog Logging Best Practices - Datadog - Practical guide on log levels, structured fields, and retention policies; reinforces the schema-design choices in this chapter.
-
Grafana Loki Documentation - Grafana Labs - Reference for the open-source log aggregation system that pairs well with this textbook's per-request JSON log schema for high-cardinality LLM logs.