scuffle-batching 0.1.2

Optimized batching and dataloading for external services.
Documentation 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109

use std::future::Future;
use std::marker::PhantomData;
use std::sync::Arc;

use criterion::{BenchmarkId, Criterion, black_box, criterion_group, criterion_main};
use scuffle_batching::batch::BatchResponse;

struct DataloaderImpl<F, P>(F, PhantomData<P>);

impl<F, P> DataloaderImpl<F, P> {
    fn new(f: F) -> Self {
        Self(f, PhantomData)
    }
}

impl<F, Fut> scuffle_batching::BatchExecutor for DataloaderImpl<F, Fut>
where
    F: Fn(Vec<(usize, BatchResponse<usize>)>) -> Fut + Send + Sync,
    Fut: Future<Output = ()> + Send,
    Self: Send + Sync,
{
    type Request = usize;
    type Response = usize;

    async fn execute(&self, keys: Vec<(Self::Request, BatchResponse<Self::Request>)>) {
        (self.0)(keys).await;
    }
}

async fn run_scuffle_batcher_many(
    size: usize,
    loader: impl scuffle_batching::BatchExecutor<Request = usize, Response = usize> + Send + Sync + 'static,
) {
    let batcher = Arc::new(scuffle_batching::Batcher::new(
        loader,
        size / 2,
        100,
        std::time::Duration::from_millis(5),
    ));

    let spawn = || {
        let batcher = batcher.clone();
        tokio::spawn(async move { batcher.execute_many(0..size / 4).await })
    };

    futures::future::join_all([spawn(), spawn(), spawn(), spawn()]).await;
}

async fn run_scuffle_batcher_single(
    size: usize,
    loader: impl scuffle_batching::BatchExecutor<Request = usize, Response = usize> + Send + Sync + 'static,
) {
    let batcher = Arc::new(scuffle_batching::Batcher::new(
        loader,
        size / 2,
        100,
        std::time::Duration::from_millis(5),
    ));

    let spawn = |i| {
        let batcher = batcher.clone();
        tokio::spawn(async move { batcher.execute(i).await })
    };

    futures::future::join_all((0..size / 4).cycle().take(size).map(spawn)).await;
}

fn delay(c: &mut Criterion) {
    let size: usize = 1000;

    let mut group = c.benchmark_group("batcher - delay");

    let runtime = || tokio::runtime::Builder::new_current_thread().enable_time().build().unwrap();

    group.bench_with_input(BenchmarkId::new("many", size), &size, |b, &s| {
        b.to_async(runtime()).iter(|| async move {
            run_scuffle_batcher_many(
                s,
                DataloaderImpl::new(|keys: Vec<(usize, BatchResponse<usize>)>| async move {
                    black_box(tokio::time::sleep(std::time::Duration::from_millis(1))).await;
                    for (key, resp) in keys {
                        resp.send(black_box(key));
                    }
                }),
            )
            .await;
        });
    });

    group.bench_with_input(BenchmarkId::new("single", size), &size, |b, &s| {
        b.to_async(runtime()).iter(|| async move {
            run_scuffle_batcher_single(
                s,
                DataloaderImpl::new(|keys: Vec<(usize, BatchResponse<usize>)>| async move {
                    black_box(tokio::time::sleep(std::time::Duration::from_millis(1))).await;
                    for (key, resp) in keys {
                        resp.send(black_box(key));
                    }
                }),
            )
            .await;
        });
    });

    group.finish();
}

criterion_group!(benches, delay);
criterion_main!(benches);