Partitioning

TidesDB tables can be partitioned with MariaDB’s standard partitioning syntax. Each partition becomes a separate TidesDB table, and therefore a separate column family, so compaction and flushes happen independently per partition.

CREATE TABLE metrics (
  id    INT NOT NULL,
  ts    DATE NOT NULL,
  value DOUBLE,
  PRIMARY KEY (id, ts)
) ENGINE=TIDESDB
PARTITION BY RANGE COLUMNS(ts) (
  PARTITION p_2024   VALUES LESS THAN ('2025-01-01'),
  PARTITION p_2025   VALUES LESS THAN ('2026-01-01'),
  PARTITION p_future VALUES LESS THAN MAXVALUE
);

All of MariaDB’s partitioning schemes work, HASH, KEY, RANGE, LIST, and RANGE COLUMNS. Secondary indexes on partitioned tables work too, with each partition holding its own index column family.

Partitions are added and dropped with ALTER TABLE. Dropping a partition removes all of its data:

ALTER TABLE metrics ADD PARTITION (PARTITION p_2026 VALUES LESS THAN ('2027-01-01'));
ALTER TABLE metrics DROP PARTITION p_2024;

One limitation to note, a partitioned table cannot carry a vector index, which is covered in Vector Search.