Kubernetes based Micro Distributed Batch Framework and Load Testing Library for C#.
This library allows you to write distributed batch or load test scenarios in C#. In addition to HTTP/1, you can test HTTP/2, gRPC, MagicOnion, Photon, or original network transport by writing in C#.
Work In Progress Preview 0.0.1, currently you can try standalone batch framework and load testing like ApacheBanch.
Install via NuGet
Install-Package DFrame
Install-Package DFrame.LoadTesting
Sample code of HTTP/1 load testing.
Implement Worker class in your assembly.
public class SampleHttpWorker : Worker
{
const string url = "http://localhost:5000";
HttpClient httpClient;
public override async Task SetupAsync(WorkerContext context)
{
var handler = new HttpClientHandler
{
MaxConnectionsPerServer = 100,
};
httpClient = new HttpClient(handler);
httpClient.DefaultRequestHeaders.Add("ContentType", "application/json");
}
public override async Task ExecuteAsync(WorkerContext context)
{
await httpClient.GetAsync(_url, cts.Token);
}
public override async Task TeardownAsync(WorkerContext context)
{
}
}Setup entrypoint RunDFrameAsync or RunDFrameLoadTestingAsync.
class Program
{
static async Task Main(string[] args)
{
await Host.CreateDefaultBuilder(args)
// .RunDFrameAsync(args, new DFrameOptions("localhost", 12345));
.RunDFrameLoadTestingAsync(args, new DFrameOptions("localhost", 12345));
}And execute commandline.
Usage: <Command> <Args>
Commands:
batch
request
rampup
batch Options:
-workerName <String> (Required)
-processCount <Int32> (Default: 1)
request Options:
-workerName <String> (Required)
-processCount <Int32> (Required)
-workerPerProcess <Int32> (Required)
-executePerWorker <Int32> (Required)
rampup Options:
-workerName <String> (Required)
-processCount <Int32> (Required)
-maxWorkerPerProcess <Int32> (Required)
-workerSpawnCount <Int32> (Required)
-workerSpawnSecond <Int32> (Required)
- processCount - scaling process count. in kuberenetes means Pod count. for inprocess, recommend to use 1.
- workerPerProcess - worker count of process. This is similar to concurrent count.
- executePerWorker - execute count per worker, if use for batch, recommend to set 1.
- workerName - execute worker name, default is type name of Worker.
SampleApp.exe "reqeust -processCount 1 -workerPerProcess 10 -executePerWorker 1000 -workerName "SampleHttpWorker"
If use RunDFrameLoadTestingAsync, shows execution result like apache bench.
Show Load Testing result report.
Finished 10000 requests
Scaling Type: InProcessScalingProvider
Request count: 10000
ProcessCount: 1
WorkerPerProcess: 10
ExecutePerWorker: 1000
Concurrency level: 10
Complete requests: 10000
Failed requests: 0
Time taken for tests: 19.15 seconds
Requests per seconds: 522.16 [#/sec] (mean)
Time per request: 19.15 [ms] (mean)
Time per request: 1.92 [ms] (mean, across all concurrent requests)
Percentage of the requests served within a certain time (ms)
50% 18
66% 19
75% 19
80% 20
90% 20
95% 20
98% 21
99% 22
100% 378 (longest request)
Data can be shared between workers via DistirbutedColleciton.
DistributedList<T>DistributedQueue<T>DistributedStack<T>DistributedHashSet<T>DistributedDictionary<TKey, TValue>
public class SampleWorker : Worker
{
IDistributedQueue<int> queue;
Random rand;
public override async Task SetupAsync(WorkerContext context)
{
queue = context.CreateDistributedQueue<int>("sampleworker-testq");
rand = new Random();
}
public override async Task ExecuteAsync(WorkerContext context)
{
await queue.EnqueueAsync(rand.Next());
}
public override async Task TeardownAsync(WorkerContext context)
{
while (true)
{
var v = await queue.TryDequeueAsync();
if (v.HasValue)
{
Console.WriteLine($"Dequeue all from {Environment.MachineName} {context.WorkerId}: {v.Value}");
}
else
{
return;
}
}
}
}WIP, DFrame scales 1-10000 workers via Kuberenetes. Distributed batches can be written collaboratively through inter-worker communication through Distributed Collections. It also enables large scale load testing.
This library is under the MIT License.