Data platform · Quality & Contracts

Schema evolution

Mediumschema evolutioncompatibilitycontractsmigrations

Interview question

What is schema evolution, and how do you handle it safely?

How to answer

The short answer

Schema evolution is how table or event schemas change over time: new columns, dropped fields, type changes, renamed keys.

Output
v1 event: {id, amount}
v2 event: {id, amount, currency}   # additive, usually safe
v3 event: {id, amount_cents}       # rename/type change, dangerous

Safer practices

  • Prefer additive changes (new optional columns)
  • Use schema registries for Kafka (compat rules: backward/forward)
  • In lakes, table formats can evolve metadata carefully
  • Version contracts; communicate breaking changes
  • Make consumers tolerant of unknown fields when possible

Breaking changes need a migration plan

Dual-write, dual-read, , then cut over.

Interview tip: Classify additive vs breaking, then mention compatibility modes and coordinated migrations.