Timeseries on Databricks Lakebase
Introducing LakeTS: time-series capabilities, now native to Lakebase.
Timeseries has usually meant standing up a separate database. LakeTS chanaes that - a pure-SOL toolkit that brings full time-series power to the Databricks Data Intelligence Platform: hot on Lakebase, governed by Unity Catalog on Cold Layer, zero extensions
What's inside:
-> ChronoTables - time-partitioned tables with pre-created chunks, BRIN indexes. and millisecond chunk drops
-> Native SQL functions - time_bucket, locf, gapfill, rate and more in pure PL/pgSQL
-> Incremental RollUps - watermarked, cascading hierarchical aggregates
-> Last Value Cache- sub-10ms reads on the latest value per key
-> Hot/cold tiering - Lakebase CDF streams older data into a Unity Catalog Managed Table for cheap, long-horizon retention
-> SQL-native alerts + bulk ingest from edge devices