What are time travel queries?

A time travel query reads a table as it existed at a past point in time or snapshot, instead of its current state. A table format like Apache Iceberg keeps every snapshot's metadata, so you can query yesterday's version of a table the same way you'd query today's.

How it works

Every time a table is written to, the table format records a new snapshot: the exact set of files that made up the table at that moment. A time travel query points at a specific snapshot, either by ID or by timestamp, and reads only the files that snapshot references, ignoring anything written after it.

Why it matters

Time travel turns "what did this look like before that change" from a question you'd need a backup for into a query you can run directly. It's useful for debugging a bad pipeline run, auditing what a report showed on a given date, or comparing two versions of the same table.

How OptimaFlo implements this

This is a live, user-facing feature, not just something Iceberg's storage layer makes theoretically possible. Every table's detail page has a time travel tab: pick a snapshot from the picker, write a query, and run it against that snapshot through DuckDB, PyIceberg, or BigQuery, whichever engine the table uses.

Related terms

• Apache Iceberg: The table format whose snapshot metadata makes time travel queries possible.
• Schema evolution: A related Iceberg capability: a table's schema can change between snapshots too.