A new column appears in incoming data. Which format handles schema evolution well?
Why B is the answer
Apache Iceberg tables
Table formats such as Iceberg support schema evolution.
In detail
- Iceberg tables support schema evolution.
- Table formats track schema as metadata.
Use case
An illustrative scenario, not a real incident.
A new column appears in incoming data. Iceberg tables add it with no rewrite of old files.
See it in different ways
stateDiagram-v2 [*] --> RawCsv RawCsv --> Parquet: columnar, compressed Parquet --> Partitioned: by date Partitioned --> Compacted: fewer small files Compacted --> [*]
stateDiagram-v2 state "Schema version 1" as s0 state "Schema version 2" as s1 state "Old and new files readable" as s2 [*] --> s0 s0 --> s1: add column s1 --> s2 s2 --> [*]
Key point: Table formats handle schema evolution.
Why the other options are wrong
- A. Plain JSON files with no catalog
- Files with no catalog do not manage a schema.
- C. CSV files with no header
- CSV with no header has no schema.
- D. Fixed-width text files
- Fixed-width files break when columns change.