Artemis is a library which dynamically resize the amount of workers (goroutines) to run functions in parallel at your target rate.
$ go get -u github.com/RY-2718/artemispackage main
import (
"context"
"fmt"
"runtime"
"time"
"github.com/RY-2718/artemis"
)
func main() {
// Define the function to run
f := func() {
time.Sleep(300 * time.Millisecond)
}
// Define the rate (frequency, error rate)
rate := artemis.Rate{
Freq: 100,
Per: time.Second, // Set `Per` param as time.Second now
ErrorRate: 0.1,
}
ctx := context.Background()
// Runner has responsibility for running function and reporting stats
runner := artemis.Runner{}
go runner.RunWithRate(ctx, &rate, f)
// If you want to report periodically, using Ticker is a solution
ticker := time.NewTicker(1 * time.Second)
for {
select {
case <-ticker.C:
// runner.Report() returns a struct which have RPS (rate per second), TargetRPS
r := runner.Report()
fmt.Printf("%f, %f, %d\n", r.RPS, r.TargetRPS, runtime.NumGoroutine())
}
}
}Here we show the performance of Artemis at [100, 1000] rps (rate per second).
Artemis increases workers exponentially at first, and decreases them by half when rps exceeds target rps, then increases again when rps gets below the target.
MIT