Skip to content

Latest commit

 

History

History
201 lines (136 loc) · 7.79 KB

File metadata and controls

201 lines (136 loc) · 7.79 KB

.NET CORE Client for Apache ActiveMQ Artemis

Build


ArtemisNetCoreClient

NuGet Status
ArtemisNetCoreClient NuGet

Apache ActiveMQ Artemis is an open-source project to build a multi-protocol, embeddable, very high performance, clustered, asynchronous messaging system.

This .NET client library is an open-source effort to equip .NET developers with a powerful, straightforward client for Apache ActiveMQ Artemis. Utilizing the broker's CORE protocol, this library focuses on high-performance messaging, ensuring compatibility and comprehensive feature support with Apache ActiveMQ Artemis.

Quickstart

Add ArtemisNetCoreClient NuGet package to your project using dotnet CLI:

dotnet add package ArtemisNetCoreClient --prerelease

The API interfaces and classes are defined in the ActiveMQ.Artemis.Core.Client namespace. Make sure to include it at the top of your file:

using ActiveMQ.Artemis.Core.Client;

Before sending or receiving messages, you need to establish a connection to the broker endpoint. Create a connection using the ConnectionFactory object.

var connectionFactory = new ConnectionFactory();
var endpoint = new Endpoint
{
    Host = "localhost",
    Port = 61616,
    User = "guest",
    Password = "guest"
};
var connection = await connectionFactory.CreateAsync(endpoint);

Once the connection is established, create a session.

var session = await connection.CreateSessionAsync();

Create an address and a queue for messaging.

await session.CreateAddressAsync("a1", [RoutingType.Anycast]);
await session.CreateQueueAsync(new QueueConfiguration
{
    Address = "a1",
    Name = "q1",
    RoutingType = RoutingType.Anycast
});

Create a producer to send a message to the address.

var producer = await session.CreateProducerAsync(new ProducerConfiguration
{
    Address = "a1"
});
await producer.SendMessageAsync(new Message { Body = "my-msg"u8.ToArray() });

Create a consumer to receive messages from the queue.

var consumer = await session.CreateConsumerAsync(new ConsumerConfiguration
{
    QueueName = "q1"
});
var message = await consumer.ReceiveMessageAsync();

Online resources

Performance

Note

The benchmark results seem to be a bit too good to be true. Please take them with a pinch of salt. Further investigation into potential misconfigurations is advised. For more details, see Issue #116.

All benchmarks are executed on the following hardware and software configuration:

OS=macOS 14.5 (23F79) [Darwin 23.5.0]
Apple M1 Pro, 1 CPU, 10 logical and 10 physical cores
.NET SDK=8.0.100
  [Host]     : .NET 8.0.0 (8.0.23.53103), Arm64 RyuJIT
  DefaultJob : .NET 8.0.0 (8.0.23.53103), Arm64 RyuJIT

Both the benchmark applications and the broker are hosted on the same machine, with the broker running inside a Docker container. The configuration for the queues used in the benchmarks is specified within the broker.xml file. Each message used in the benchmarks contains a simple payload consisting of 1KB of randomly generated data.

The baseline for these benchmarks is set using the official Apache NMS.AMQP library.

Throughput

This benchmark assesses the ability of the library to handle large volumes of messages efficiently, focusing on both sending and receiving processes:

  • Producer: Sends 100,000 messages. The performance metric is the rate of messages sent per second (msgs/s).
  • Consumer: Receives an equivalent volume of messages. The performance metric is the rate of messages processed per second (msgs/s).

To ensure that broker disk I/O does not influence the results, messages are sent as non-durable.

Results show how quickly each library can transmit and receive messages:

Library Sending msgs/s Consuming msgs/s
ArtemisNetCoreClient 135,550.65 138,047.69
NMS.AMQP 67,937.65 65,838.88
Benchmark Results Diagram
Benchmark Results Diagram

Request-Reply (PingPong) Benchmark

This benchmark measures the performance in a request-reply messaging scenario. The test setup involves two components, Ping and Pong:

  • Ping: Sends a message to Pong and starts a timer.
  • Pong: Receives the message and responds back immediately.

The performance is assessed by measuring the round-trip time and calculating the throughput in terms of round-trips per second, over the course of sending and receiving 10,000 messages.

Library Round-trips per Second
ArtemisNetCoreClient 2,205.86
NMS.AMQP 905.55
Benchmark Results Diagram

Latency Benchmark

This benchmark measures the latency of message delivery between a producer and a consumer.

  • Producer: Initiates the sending of 10,000 messages, each embedded with a high-precision timestamp immediately before dispatch.
  • Consumer: Receives messages and calculates the latency for each by comparing the embedded timestamp with the time of receipt.
Library Average Latency (µs)
ArtemisNetCoreClient 524.31
NMS.AMQP 552.49
Benchmark Results Diagram

Running the tests

To run the tests, you need an Apache ActiveMQ Artemis server. The server can be hosted in a Docker container.

Setting Up the Necessary Infrastructure

Ensure that Docker and Docker Compose are properly installed and configured on your machine.

  1. Navigate to the /test/artemis directory.
  2. Run the following command to spin up the broker:
docker-compose up -V -d

With the broker up and running, you can execute the test suite using the following command:

dotnet test

Disclaimer

Please note that this project is currently under active development and is not considered production-ready. We are continuously working to improve and stabilize its features, but it does not yet meet all the requirements for production use.

If you are in search of a production-ready ActiveMQ Artemis client for .NET, we recommend checking out the AMQP-based client. This alternative client has been battle-tested in production environments for the last few years and supports a wide range of ActiveMQ Artemis features.

License

This project is licensed under the Apache-2.0 License. You are welcome to use it freely, without any restrictive obligations.

Footnotes

  1. Examples in this article were created using AMQP-based client.