#async #futures #ratelimit #throttle #tokenbucket


A token-based rate limiter based on the leaky bucket algorithm

26 releases (1 stable)

1.0.0 May 12, 2023
0.12.5 May 12, 2023
0.12.4 Mar 22, 2023
0.12.1 Jul 10, 2022
0.4.1 Oct 6, 2019

#24 in Concurrency

Download history 3459/week @ 2023-02-03 3768/week @ 2023-02-10 3192/week @ 2023-02-17 3970/week @ 2023-02-24 3977/week @ 2023-03-03 4315/week @ 2023-03-10 5008/week @ 2023-03-17 4164/week @ 2023-03-24 4923/week @ 2023-03-31 5126/week @ 2023-04-07 4814/week @ 2023-04-14 5889/week @ 2023-04-21 5459/week @ 2023-04-28 6255/week @ 2023-05-05 5654/week @ 2023-05-12 4526/week @ 2023-05-19

23,212 downloads per month
Used in 17 crates (9 directly)


750 lines


github crates.io docs.rs build status

A token-based rate limiter based on the leaky bucket algorithm.

If the bucket overflows and goes over its max configured capacity, the task that tried to acquire the tokens will be suspended until the required number of tokens has been drained from the bucket.

Since this crate uses timing facilities from tokio it has to be used within a Tokio runtime with the time feature enabled.

This library has some neat features, which includes:

  • Not requiring a background thread or task to drip. Instead, one of the waiting tasks assumes the duty as coordinator called the core. See below for more details.
  • Dropped tasks releases their partially acquired tokens.


The core type is the RateLimiter type, which allows for limiting the throughput of a section using its acquire and acquire_one methods.

use leaky_bucket::RateLimiter;
use std::time;

async fn main() {
    let limiter = RateLimiter::builder()

    let start = time::Instant::now();

    println!("Waiting for permit...");

    // Should take ~400 ms to acquire in total.
    let a = limiter.acquire(7);
    let b = limiter.acquire(3);
    let c = limiter.acquire(10);

    let ((), (), ()) = tokio::join!(a, b, c);

        "I made it in {:?}!",

Implementation details

Each rate limiter has two acquisition modes. A fast path and a slow path. The fast path is used if the desired number of tokens are readily available, and involves incrementing an atomic counter indicating that the acquired number of tokens have been added to the bucket.

If this counter goes over its configured maximum capacity, it overflows into a slow path. Here one of the acquiring tasks will switch over to work as a core. This is known as core switching.

use leaky_bucket::RateLimiter;
use std::time;

let limiter = RateLimiter::builder()

// This is instantaneous since the rate limiter starts with 10 tokens to
// spare.

// This however needs to core switch and wait for a while until the desired
// number of tokens is available.

The core is responsible for sleeping for the configured interval so that more tokens can be added. After which it ensures that any tasks that are waiting to acquire including itself are appropriately unsuspended.

On-demand core switching is what allows this rate limiter implementation to work without a coordinating background thread. But we need to ensure that any asynchronous tasks that uses RateLimiter must either run an acquire call to completion, or be cancelled by being dropped.

If none of these hold, the core might leak and be locked indefinitely preventing any future use of the rate limiter from making progress. This is similar to if you would lock an asynchronous Mutex but never drop its guard.

You can run this example with:

cargo run --example block-forever
use leaky_bucket::RateLimiter;
use std::future::Future;
use std::sync::Arc;
use std::task::Context;

struct Waker;

let limiter = Arc::new(RateLimiter::builder().build());

let waker = Arc::new(Waker).into();
let mut cx = Context::from_waker(&waker);

let mut a0 = Box::pin(limiter.acquire(1));
// Poll once to ensure that the core task is assigned.
assert!(a0.as_mut().poll(&mut cx).is_pending());

// We leak the core task, preventing the rate limiter from making progress
// by assigning new core tasks.

// Awaiting acquire here would block forever.
// limiter.acquire(1).await;


By default RateLimiter uses a fair scheduler. This ensures that the core task makes progress even if there are many tasks waiting to acquire tokens. As a result it causes more frequent core switching, increasing the total work needed. An unfair scheduler is expected to do a bit less work under contention. But without fair scheduling some tasks might end up taking longer to acquire than expected.

This behavior can be tweaked with the Builder::fair option.

use leaky_bucket::RateLimiter;

let limiter = RateLimiter::builder()

The unfair-scheduling example can showcase this phenomenon.

cargh run --example unfair-scheduling
# fair
Max: 1011ms, Total: 1012ms
 0: 101ms
 1: 101ms
 2: 101ms
 3: 101ms
 4: 101ms
# unfair
Max: 1014ms, Total: 1014ms
 0: 1014ms
 1: 101ms
 2: 101ms
 3: 101ms
 4: 101ms

As can be seen above the first task in the unfair scheduler takes longer to run because it prioritises releasing other tasks waiting to acquire over itself.


~128K SLoC