Course contentsShow
← Software Engineering
Lesson 1131 of 1,93334. Logging and ObservabilityFree lesson

JSON and Other Structured Formats

Compare structured logging formats including JSON, logfmt, and how to choose the right format for your needs.

JSON and Other Structured Formats

You'll learn: How to choose between JSON, logfmt, and other structured logging formats based on your system's needs.

Understanding Structured Format Options

Once you've decided to use structured logging with key-value pairs, you need to pick a serialization format. The two most common choices are JSON and logfmt, each with distinct trade-offs.

JSON Format

JSON is the most widely adopted structured logging format:

{"timestamp":"2024-01-15T10:30:45Z","level":"error","user_id":"u123","action":"checkout","error":"payment_failed","amount":99.99}

Advantages:

  • Universal tooling support (every language can parse it)
  • Rich data types (nested objects, arrays, booleans, nulls)
  • Easy integration with log aggregation systems
  • Machine-first format optimized for parsing

Disadvantages:

  • Harder for humans to read in raw form
  • More verbose (larger log files)
  • Slower to parse than simpler formats

Logfmt Format

Logfmt is a simpler, human-readable alternative:

timestamp=2024-01-15T10:30:45Z level=error user_id=u123 action=checkout error=payment_failed amount=99.99

Advantages:

  • Extremely human-readable
  • Faster parsing
  • More compact (less storage/bandwidth)
  • Great for local development

Disadvantages:

  • Limited data types (everything is strings)
  • No nested structures
  • Less tooling support than JSON

Choosing the Right Format

Use JSON when you need rich data types, nested structures, or integration with centralized logging systems. Use logfmt for human-readability during development or when log volume and parsing speed matter most.

Key Takeaway: JSON offers power and compatibility at the cost of verbosity, while logfmt provides simplicity and readability—choose based on whether your logs are primarily for machines or humans.