Bintrail adds time‑travel and row‑history queries to MySQL via indexed binlogs

Bintrail is a new layer that brings point‑in‑time and row‑history queries to MySQL by indexing ROW‑format binary logs and routing historical queries through ProxySQL. That capability gives teams SQL access to past table and row states — making targeted recovery and forensic queries possible without full‑restore workflows.

Under the hood, Bintrail parses ROW‑format binlogs and indexes every row event with full before‑and‑after images. It can generate reversal SQL for point‑in‑time recovery even when the original binlog files are not present, and exposes SQL constructs such as _flashback and _diff for querying history. Example usage in the project documentation shows queries like: SELECT * FROM _flashback.orders AS OF '2026 — 04-15 09:30:00' WHERE id = 42; and a generic SELECT * FROM _diff for change comparisons.

The layer sits behind ProxySQL and uses query routing so normal MySQL traffic remains untouched while historical queries are directed to Bintrail’s backend. Bintrail can automatically create ProxySQL routing rules for patterns such as _flashback, _diff, and _snapshot. It keeps an independent indexed history store that is not bound to MySQL’s binlog retention policy and can optionally extend lookups into archived Parquet files stored on S3.

Bintrail is positioned to fill a long‑standing MySQL gap: most other major OLTP systems offer queryable history or temporal tables natively. Oracle supports AS OF TIMESTAMP, SQL Server provides FOR SYSTEM_TIME AS OF, MariaDB ships system‑versioned tables, PostgreSQL users rely on extensions, and CockroachDB included time travel from its first release. According to project author and maintainer Daniel Guzmán Burgos, MySQL historically left this capability to operational binlog tooling rather than SQL.

For builders and operations teams, the practical benefits are concrete. Bintrail can reconstruct the full sequence of changes for a given row-including GTID and event type-return before/after images, and produce SQL to revert or replay state. Those features reduce reliance on heavy full‑backup restores for single‑row or point‑in‑time repairs and enable longer historical queries than native binlog windows permit.

Project maintainers and community observers have framed Bintrail as a pragmatic response to that gap. Daniel Guzmán Burgos described mapping how every major OLTP except MySQL offered point‑in‑time queries as the motivation; Peter Zaitsev of Percona praised the work for improving recoverability; and Roman Agabekov of Releem said full backup restores are often too slow or risky, underscoring the operational appeal of lighter‑weight time‑travel tooling.