Experience Database
Like Never Before

Get sub-millisecond performance on analytical workloads through columnar storage and SIMD vectorization — all from safe, zero-copy Rust.

Get Started View on GitHub

Zero Overhead

Calls the RayforceDB core's C API directly — no marshalling shim. Reads are zero-copy: a numeric column is a &[T] slice, not a per-element copy.

Performance

Columnar storage and SIMD vectorization deliver consistent sub-millisecond latency on analytical workloads. Build a column in a single memcpy; read it back as a borrow.

Ecosystem

RayforceDB spans multiple languages and in-house tools, from the C core to editor tooling — enabling high-performance, distributed solutions.

Fluent Queries

Chainable query builders that read like the operation. Operator overloads for arithmetic, methods for comparisons, aggregations either way.

Type-Safe & Zero-Copy

A full value model — atoms, vectors, lists, dicts, tables — with ToValue/FromValue conversions and optional chrono temporals.

TCP Servers

Execute queries over TCP with low latency. Connect to servers, organize data in a distributed manner, and build distributed processing applications.

Rayforce-RS in a few lines

Build a table, run a grouped aggregation, and print it — all from safe Rust.

use rayforce::{col, sum, Runtime, Table, Value};

Runtime::scope(|_rt| {
    let t = Table::new(
        &["sym", "price", "size"],
        &[
            Value::sym_vec(&["AAPL", "MSFT", "AAPL", "GOOG"]),
            Value::vec(&[100.0f64, 200.0, 110.0, 300.0]),
            Value::vec(&[10i64, 20, 30, 40]),
        ],
    )?;

    let totals = t
        .select()
        .agg("total", sum(col("size")))
        .filter(col("price").gt(150.0))
        .by("sym")
        .execute()?;

    println!("{totals}");
    Ok(())
})?;