Large Amount of Column Families with RocksDB and TidesDB
published on February 5th, 2026
Just scouring the internet during some researching I was doing, I came a known issue with RocksDB when performing operations across many column families.
See: RocksDB issue #5117
The plot was created by modifying the Java program at https://github.com/grove/bugs-and-regressions/tree/master to compare RocksDB against TidesDB.
Environment
- Intel Core i7-11700K (8 cores, 16 threads) @ 4.9GHz
- 48GB DDR4
- Western Digital 500GB WD Blue 3D NAND Internal PC SSD (SATA)
- Ubuntu 23.04 x86_64 6.2.0-39-generic
- TidesDB v8.1.2
- RocksDB v10.10.1
- GCC (glibc)
This is running on the latest version of TidesDB and RocksDB. Thus the issue from many years ago still lingers in RocksDB.
TidesDB shows strong performance when working across a large number of column families. This is something I rely on heavily in the MariaDB pluggable engine, TideSQL, where column families are used extensively. In a relational system, I store metadata, indexes, and more using this approach. Building a pluggable storage engine has not been easy.
MySQL and now MariaDB provide a wonderful interface to work with. It is intuitive to get started and then expand on what you learn while wiring things together.
I like to think of it as tying an engine to a car or a plane. There are many wires, connections, processes, and safety considerations, and I enjoy it.
I look forward to seeing what you build with TidesDB!
Code for comparison:
Raw data: