Will the Guru

For message to be transferred, it needs a process called ‘serialization’ and ‘deserialization’.

---

Data Formats

1. Textual formats

• ex) JSON, XML, CSV, …
• Pros
  • Human-readable (easier to debug and test)
  • Widely supported by languages and tools
• Cons
  • Bigger messages (slower to transfer)
  • Slower serialization and deserialization

2. Binary formats

• ex) Protobuf, Thrift, Avro, …
• Utilizes schema to serialize/deserialize
• Pros
  • Smaller messages (faster to transfer and less space needed to store)
    • ex) Integers are encoded into a single byte instead of several bytes
  • Faster to serialize/deserialize
• Cons
  • Not human-readable (harder to debug and test)

Backward compatibility is important for distributed systems.

---

Reference

• https://leetcode.com/explore/interview/card/system-design-for-interviews-and-beyond/724/fundamentals-of-reliable-scalable-and-fast-communi/4757/