RayforceDB
Comment: Most of the sources are either not reliable such as WP:user-generated sources (blogs, github, etc.), Technewsreel is essentially a regurgitation of their website with no named author and similar for RepoRank, Brevfeed, etc. (no editorial oversight). S0091 (talk) 17:10, 30 September 2026 (UTC)
| File:RayforceDB Logo.svg | |
| Original author(s) | Anton Kundenko |
|---|---|
| Initial release | 2025 (open-source release) |
| Stable release | 2.10.0
/ September 29, 2026 |
| Repository | github |
| Written in | C |
| Engine | |
| Operating system | Linux, macOS |
| Type | Column-oriented DBMS, graph database, embedded database |
| License | MIT License |
| Website | rayforcedb |
Search RayforceDB on Amazon.
RayforceDB (also written Rayforce) is an open-source columnar analytics database engine written in the C programming language. It combines columnar query processing with graph traversal and vector similarity search in a single execution engine, and it is queried through Rayfall, a Lisp-like query language, rather than SQL.[1] The engine is designed to be embedded in applications for low-latency analytical workloads, and it has been compared with and DuckDB.[1][2][3] The software is distributed under the MIT License and can be used as a library or interactively through a read–eval–print loop.[1][4]
History
Development of the engine began in 2023. The project originates from Ukraine, Its principal author is the software engineer Anton Kundenko.[4][5] The source code was published on GitHub under the MIT License, and by April 2026 the project was described as recently open-sourced.[2] Version 2.10.0 was released on 29 September 2026.[6]
Features
Execution engine
RayforceDB is written in C with no external dependencies beyond the C standard library.[1][3] Queries are represented as a lazy directed acyclic graph of operations covering scans, filters, joins, aggregations, window functions and graph traversals; a multi-pass optimizer rewrites the graph before execution, applying steps such as constant folding, predicate pushdown, projection pushdown and partition pruning.[7] Element-wise parts of a query are compiled to bytecode at execution time and evaluated over fixed-size batches of rows, an approach the project describes as morsel-driven execution.[7] The engine replaces the system memory allocator with a buddy allocator and per-thread heaps.[1][8]
Graph and vector search
Graph data is stored in compressed sparse row form, and the engine implements traversal and analysis algorithms including breadth-first search, Dijkstra's algorithm, PageRank and community detection.[1][8] Approximate nearest neighbor search over vector data is provided through a Hierarchical navigable small world (HNSW) index.[1] The project also supports recursive queries in a Datalog style.[5]
Storage
Tables are stored as columnar files that can be memory-mapped, including splayed and date-partitioned layouts, and a CSV reader with type inference is included.[8][7] A reviewer noted that string columns are automatically stored as enumerations backed by shared symbol files.
Rayfall query language
Rather than SQL, RayforceDB is queried with Rayfall, a Lisp-like language that uses prefix notation and S-expressions.[1] The following example, taken from the project documentation, creates a table and aggregates it by one column:[8]
(set t (table [Symbol Side Qty]
(list [AAPL GOOG MSFT AAPL GOOG]
[Buy Sell Buy Sell Buy]
[100 200 150 300 250])))
(select {from:t by: Symbol Qty: (sum Qty)})
Lambdas written in Rayfall are compiled lazily to bytecode and run in a virtual machine inside the engine.[8]
Language support
The engine can be used through the interactive Rayfall REPL, through a C API exposed by a single header file, and through client bindings for Python and Rust.[4][8] Prebuilt packages are provided for Linux and macOS, including a Homebrew formula and a Debian package; Windows is not among the supported platforms.[1][3][8]
Use in other projects
PeachQ, an MIT-licensed open-source implementation of the q programming language, uses Rayforce as its in-memory engine for vector operations, tables and interactive analysis, with DuckDB-backed storage planned for later releases.[9][10]
Reception
In an April 2026 review, quantitative-finance author Anton Vorobets reported that some intraday time-series workloads that were impractical in SQL databases completed in under 30 seconds using RayforceDB's Python bindings.[2] He also noted that moving from SQL to Rayfall requires a change of habits and an initial learning investment.[2]
A profile on the repository-analysis site RepoRank characterised RayforceDB as a specialist alternative to DuckDB and ClickHouse, recommending those systems where a broader ecosystem is needed, and noted that platform support was limited to x86-64 Linux and ARM64 macOS.[3]
The project reports use in production systems in the financial sector.[5][11]
References
- ↑ 1.0 1.1 1.2 1.3 1.4 1.5 1.6 1.7 1.8 "RayforceDB Unifies Columnar Analytics and Graph Engines in Pure C". TechNewsReel. 14 August 2026. Retrieved 30 September 2026.
- ↑ 2.0 2.1 2.2 2.3 Vorobets, Anton (16 April 2026). "Time Series Database Review: RayforceDB". Substack. Retrieved 30 September 2026.
- ↑ 3.0 3.1 3.2 3.3 "Rayforce: SIMD-accelerated Columnar Analytics Database". RepoRank. 8 July 2026. Retrieved 30 September 2026.
- ↑ 4.0 4.1 4.2 "RayforceDB". Database of Databases. Carnegie Mellon University. Retrieved 30 September 2026.
- ↑ 5.0 5.1 5.2 "Rayforce — Columnar Analytics and Graph Traversal in One Pipeline". rayforcedb.com. Retrieved 30 September 2026.
- ↑ "Release v2.10.0". GitHub. 29 September 2026. Retrieved 30 September 2026.
- ↑ 7.0 7.1 7.2 "Architecture: Pipeline". Rayforce documentation. Retrieved 30 September 2026.
- ↑ 8.0 8.1 8.2 8.3 8.4 8.5 8.6 "RayforceDB/rayforce". GitHub. Retrieved 30 September 2026.
- ↑ "PeachQ – open source q". peachq.org. Retrieved 30 September 2026.
- ↑ Hamilton, Ryan (29 September 2026). "TorQ loads on PeachQ 0.88". peachq.org. Retrieved 30 September 2026.
- ↑ "RayforceDB: A Pure C Analytics Database with Lisp-like Syntax for Columnar and Graph Data". BrevFeed. Retrieved 30 September 2026.
External links
This article "RayforceDB" is from Wikipedia. The list of its authors can be seen in its historical and/or the page Edithistory:RayforceDB. Articles copied from Draft Namespace on Wikipedia could be seen on the Draft Namespace of Wikipedia and not main one.
