Fast, simple, and high-performance web framework for Rust, built on top of Tokio and hyper.

Volga is designed to make building HTTP services straightforward and explicit, while keeping performance predictable and overhead minimal.

💡 Status: Volga is currently in preview.
The public API may change while core abstractions are being finalized.

Tutorial | API Docs | Examples | Roadmap

Volga focuses on clarity and control without sacrificing performance.

It avoids hidden behavior and framework-driven magic. Macros are used sparingly and primarily to reduce boilerplate. Handlers, middleware, and routing behave exactly as they look in code.

Volga is a good fit if you:

• Want simple and readable handler signatures
• Care about predictable performance and low overhead
• Need fine-grained control over the HTTP request/response lifecycle
• Work with streaming, WebSockets, or long-lived connections
• Prefer explicit APIs over code generation

• HTTP/1 and HTTP/2 support
• Explicit and robust routing
• Composable async middlewares
• Dependency Injection without derive macros
• Typed request extraction
• WebSockets and WebSocket-over-HTTP/2
• Streaming-friendly HTTP
• End-to-end OAuth 2.1/OIDC support
• Full Tokio compatibility
• Runs on stable Rust 1.90+

[dependencies]
volga = "0.9.6"
tokio = { version = "1", features = ["full"] }

use volga::{App, ok};

#[tokio::main]
async fn main() -> std::io::Result<()> {
    let mut app = App::new();
    
    app.map_get("/hello/{name}", async |name: String| {
        ok!("Hello {name}!")
    });
    
    app.run().await
}

This example demonstrates:

• typed path parameter extraction
• async request handlers
• minimal setup with zero boilerplate

More advanced examples (middleware, DI, auth, rate limiting) can be found in the documentation and here.

Volga is benchmarked using a minimal plaintext endpoint to measure raw request handling to measure baseline HTTP throughput.

The benchmark harness is available here: https://github.com/RomanEmreis/volga-benchmark

Tested on a single machine:

Platform: Apple Silicon MacBook Pro
Runtime: Linux containers (Colima)

Running 30s test @ http://volga:7878/plaintext
  8 threads and 512 connections
  Thread Stats   Avg      Stdev     Max   +/- Stdev
    Latency   539.89us    3.00ms 213.60ms   99.94%
    Req/Sec   131.04k    14.14k 296.06k    95.76%
  31080913 requests in 29.83s, 3.76GB read
  Socket errors: connect 0, read 0, write 0, timeout 987
Requests/sec: 1,041,952.40
Transfer/sec: 129.18MB

⚠️ Benchmark results are provided for reference only. Actual performance depends on workload, middleware, and handler logic.

Volga is licensed under the MIT License. Contributions welcome!