Panduan Kerangka Akson

1. Ikhtisar

Dalam artikel ini, kita akan melihat Axon dan bagaimana Axon membantu kita mengimplementasikan aplikasi dengan CQRS (Command Query Responsibility Segregation) dan Event Sourcing .

Selama panduan ini, Axon Framework dan Axon Server akan digunakan. Yang pertama akan berisi implementasi kami dan yang terakhir akan menjadi solusi Event Store dan Message Routing khusus kami.

Aplikasi sampel yang akan kami buat berfokus pada domain Order . Untuk ini, kami akan memanfaatkan blok penyusun CQRS dan Event Sourcing yang disediakan Axon .

Perhatikan bahwa banyak konsep bersama muncul langsung dari DDD, yang berada di luar cakupan artikel ini.

2. Ketergantungan Maven

Kami akan membuat aplikasi Axon / Spring Boot. Oleh karena itu, kita perlu menambahkan dependensi axon-spring-boot-starter terbaru ke pom.xml , serta dependensi axon-test untuk pengujian:

 org.axonframework axon-spring-boot-starter 4.1.2   org.axonframework axon-test 4.1.2 test 

3. Server Akson

Kami akan menggunakan Server Axon untuk menjadi Toko Acara kami dan solusi perutean perintah, acara, dan kueri khusus kami.

Sebagai Toko Acara, ini memberi kita karakteristik ideal yang diperlukan saat menyimpan acara. Artikel ini memberikan latar belakang mengapa ini diinginkan.

Sebagai solusi Perutean Pesan, ini memberi kita opsi untuk menghubungkan beberapa instans bersama-sama tanpa berfokus pada konfigurasi hal-hal seperti RabbitMQ atau topik Kafka untuk berbagi dan mengirim pesan.

Axon Server dapat diunduh di sini. Karena ini adalah file JAR sederhana, operasi berikut sudah cukup untuk memulainya:

java -jar axonserver.jar

Ini akan memulai instance Server Axon tunggal yang dapat diakses melalui localhost: 8024 . Endpoint memberikan gambaran umum tentang aplikasi yang terhubung dan pesan yang dapat mereka tangani, serta mekanisme query ke Event Store yang terdapat dalam Axon Server.

Konfigurasi default Axon Server bersama dengan ketergantungan axon-spring-boot-starter akan memastikan layanan Order kami akan secara otomatis terhubung dengannya.

4. API Layanan Pesanan - Perintah

Kami akan menyiapkan layanan Pesanan kami dengan mempertimbangkan CQRS. Oleh karena itu kami akan menekankan pesan yang mengalir melalui aplikasi kami.

Pertama, kita akan mendefinisikan Perintah, yang berarti ekspresi maksud. Layanan Order mampu menangani tiga jenis tindakan:

  1. Menempatkan pesanan baru
  2. Mengonfirmasi pesanan
  3. Mengirim pesanan

Biasanya, akan ada tiga pesan perintah yang dapat ditangani domain kita - PlaceOrderCommand , ConfirmOrderCommand , dan ShipOrderCommand :

public class PlaceOrderCommand { @TargetAggregateIdentifier private final String orderId; private final String product; // constructor, getters, equals/hashCode and toString } public class ConfirmOrderCommand { @TargetAggregateIdentifier private final String orderId; // constructor, getters, equals/hashCode and toString } public class ShipOrderCommand { @TargetAggregateIdentifier private final String orderId; // constructor, getters, equals/hashCode and toString }

The TargetAggregateIdentifier penjelasan memberitahu Axon bahwa bidang dijelaskan adalah id dari agregat diberikan kepada yang perintah harus ditargetkan. Kami akan membahas secara singkat agregat nanti di artikel ini.

Juga, perhatikan bahwa kami menandai bidang dalam perintah sebagai final. Hal ini disengaja, karena itu adalah praktek terbaik untuk setiap pelaksanaan pesan menjadi berubah .

5. Order Service API - Acara

Agregat kami akan menangani perintah , karena bertanggung jawab untuk memutuskan apakah Pesanan dapat ditempatkan, dikonfirmasi, atau dikirim.

Ini akan memberi tahu sisa penerapan keputusannya dengan menerbitkan acara. Kami akan memiliki tiga jenis peristiwa - OrderPlacedEvent, OrderConfirmedEvent , dan OrderShippedEvent :

public class OrderPlacedEvent { private final String orderId; private final String product; // default constructor, getters, equals/hashCode and toString } public class OrderConfirmedEvent { private final String orderId; // default constructor, getters, equals/hashCode and toString } public class OrderShippedEvent { private final String orderId; // default constructor, getters, equals/hashCode and toString }

6. Model Perintah - Order Agregat

Sekarang kita telah membuat model API inti kita sehubungan dengan perintah dan kejadian, kita dapat mulai membuat Model Perintah.

As our domain focuses on dealing with Orders, we'll create an OrderAggregate as the center of our Command Model.

6.1. Aggregate Class

Thus, let's create our basic aggregate class:

@Aggregate public class OrderAggregate { @AggregateIdentifier private String orderId; private boolean orderConfirmed; @CommandHandler public OrderAggregate(PlaceOrderCommand command) { AggregateLifecycle.apply(new OrderPlacedEvent(command.getOrderId(), command.getProduct())); } @EventSourcingHandler public void on(OrderPlacedEvent event) { this.orderId = event.getOrderId(); orderConfirmed = false; } protected OrderAggregate() { } }

The Aggregate annotation is an Axon Spring specific annotation marking this class as an aggregate. It will notify the framework that the required CQRS and Event Sourcing specific building blocks need to be instantiated for this OrderAggregate.

As an aggregate will handle commands which are targeted for a specific aggregate instance, we need to specify the identifier with the AggregateIdentifier annotation.

Our aggregate will commence its life cycle upon handling the PlaceOrderCommand in the OrderAggregate ‘command handling constructor'. To tell the framework that the given function is able to handle commands, we'll add the CommandHandler annotation.

When handling the PlaceOrderCommand, it will notify the rest of the application that an order was placed by publishing the OrderPlacedEvent. To publish an event from within an aggregate, we'll use AggregateLifecycle#apply(Object…).

From this point, we can actually start to incorporate Event Sourcing as the driving force to recreate an aggregate instance from its stream of events.

We start this off with the ‘aggregate creation event', the OrderPlacedEvent, which is handled in an EventSourcingHandler annotated function to set the orderId and orderConfirmed state of the Order aggregate.

Also note that to be able to source an aggregate based on its events, Axon requires a default constructor.

6.2. Aggregate Command Handlers

Now that we have our basic aggregate, we can start implementing the remaining command handlers:

@CommandHandler public void handle(ConfirmOrderCommand command) { apply(new OrderConfirmedEvent(orderId)); } @CommandHandler public void handle(ShipOrderCommand command) { if (!orderConfirmed) { throw new UnconfirmedOrderException(); } apply(new OrderShippedEvent(orderId)); } @EventSourcingHandler public void on(OrderConfirmedEvent event) { orderConfirmed = true; }

The signature of our command and event sourcing handlers simply states handle({the-command}) and on({the-event}) to maintain a concise format.

Additionally, we've defined that an Order can only be shipped if it's been confirmed. Thus, we'll throw an UnconfirmedOrderException if this is not the case.

This exemplifies the need for the OrderConfirmedEvent sourcing handler to update the orderConfirmed state to true for the Order aggregate.

7. Testing the Command Model

First, we need to set up our test by creating a FixtureConfiguration for the OrderAggregate:

private FixtureConfiguration fixture; @Before public void setUp() { fixture = new AggregateTestFixture(OrderAggregate.class); }

The first test case should cover the simplest situation. When the aggregate handles the PlaceOrderCommand, it should produce an OrderPlacedEvent:

String orderId = UUID.randomUUID().toString(); String product = "Deluxe Chair"; fixture.givenNoPriorActivity() .when(new PlaceOrderCommand(orderId, product)) .expectEvents(new OrderPlacedEvent(orderId, product));

Next, we can test the decision-making logic of only being able to ship an Order if it's been confirmed. Due to this, we have two scenarios — one where we expect an exception, and one where we expect an OrderShippedEvent.

Let's take a look at the first scenario, where we expect an exception:

String orderId = UUID.randomUUID().toString(); String product = "Deluxe Chair"; fixture.given(new OrderPlacedEvent(orderId, product)) .when(new ShipOrderCommand(orderId)) .expectException(IllegalStateException.class); 

And now the second scenario, where we expect an OrderShippedEvent:

String orderId = UUID.randomUUID().toString(); String product = "Deluxe Chair"; fixture.given(new OrderPlacedEvent(orderId, product), new OrderConfirmedEvent(orderId)) .when(new ShipOrderCommand(orderId)) .expectEvents(new OrderShippedEvent(orderId));

8. The Query Model – Event Handlers

So far, we've established our core API with the commands and events, and we have the Command model of our CQRS Order service, the Order aggregate, in place.

Next, we can start thinking of one of the Query Models our application should service.

One of these models is the OrderedProducts:

public class OrderedProduct { private final String orderId; private final String product; private OrderStatus orderStatus; public OrderedProduct(String orderId, String product) { this.orderId = orderId; this.product = product; orderStatus = OrderStatus.PLACED; } public void setOrderConfirmed() { this.orderStatus = OrderStatus.CONFIRMED; } public void setOrderShipped() { this.orderStatus = OrderStatus.SHIPPED; } // getters, equals/hashCode and toString functions } public enum OrderStatus { PLACED, CONFIRMED, SHIPPED }

We'll update this model based on the events propagating through our system. A Spring Service bean to update our model will do the trick:

@Service public class OrderedProductsEventHandler { private final Map orderedProducts = new HashMap(); @EventHandler public void on(OrderPlacedEvent event) { String orderId = event.getOrderId(); orderedProducts.put(orderId, new OrderedProduct(orderId, event.getProduct())); } // Event Handlers for OrderConfirmedEvent and OrderShippedEvent... }

As we've used the axon-spring-boot-starter dependency to initiate our Axon application, the framework will automatically scan all the beans for existing message-handling functions.

As the OrderedProductsEventHandler has EventHandler annotated functions to store an OrderedProduct and update it, this bean will be registered by the framework as a class that should receive events without requiring any configuration on our part.

9. The Query Model – Query Handlers

Next, to query this model, for example, to retrieve all the ordered products, we should first introduce a Query message to our core API:

public class FindAllOrderedProductsQuery { }

Second, we'll have to update the OrderedProductsEventHandler to be able to handle the FindAllOrderedProductsQuery:

@QueryHandler public List handle(FindAllOrderedProductsQuery query) { return new ArrayList(orderedProducts.values()); }

The QueryHandler annotated function will handle the FindAllOrderedProductsQuery and is set to return a List regardless, similarly to any ‘find all' query.

10. Putting Everything Together

We've fleshed out our core API with commands, events, and queries, and set up our Command and Query model by having an OrderAggregate and OrderedProducts model.

Next is to tie up the loose ends of our infrastructure. As we're using the axon-spring-boot-starter, this sets a lot of the required configuration automatically.

First, as we want to leverage Event Sourcing for our Aggregate, we'll need an EventStore. Axon Server which we have started up in step three will fill this hole.

Secondly, we need a mechanism to store our OrderedProduct query model. For this example, we can add h2 as an in-memory database and spring-boot-starter-data-jpa for ease of use:

 org.springframework.boot spring-boot-starter-data-jpa com.h2database h2 runtime 

10.1. Setting up a REST Endpoint

Next, we need to be able to access our application, for which we'll be leveraging a REST endpoint by adding the spring-boot-starter-web dependency:

 org.springframework.boot spring-boot-starter-web 

From our REST endpoint, we can start dispatching commands and queries:

@RestController public class OrderRestEndpoint { private final CommandGateway commandGateway; private final QueryGateway queryGateway; // Autowiring constructor and POST/GET endpoints }

The CommandGateway is used as the mechanism to send our command messages, and the QueryGateway, in turn, to send query messages. The gateways provide a simpler, more straightforward API, compared to the CommandBus and QueryBus that they connect with.

From here on, our OrderRestEndpoint should have a POST endpoint to place, confirm, and ship an order:

@PostMapping("/ship-order") public void shipOrder() { String orderId = UUID.randomUUID().toString(); commandGateway.send(new PlaceOrderCommand(orderId, "Deluxe Chair")); commandGateway.send(new ConfirmOrderCommand(orderId)); commandGateway.send(new ShipOrderCommand(orderId)); }

This rounds up the Command side of our CQRS application.

Now, all that's left is a GET endpoint to query all the OrderedProducts:

@GetMapping("/all-orders") public List findAllOrderedProducts() { return queryGateway.query(new FindAllOrderedProductsQuery(), ResponseTypes.multipleInstancesOf(OrderedProduct.class)).join(); }

In the GET endpoint, we leverage the QueryGateway to dispatch a point-to-point query. In doing so, we create a default FindAllOrderedProductsQuery, but we also need to specify the expected return type.

As we expect multiple OrderedProduct instances to be returned, we leverage the static ResponseTypes#multipleInstancesOf(Class) function. With this, we have provided a basic entrance into the Query side of our Order service.

We completed the setup, so now we can send some commands and queries through our REST Controller once we've started up the OrderApplication.

POST-ing to endpoint /ship-order will instantiate an OrderAggregate that'll publish events, which, in turn, will save/update our OrderedProducts. GET-ing from the /all-orders endpoint will publish a query message that'll be handled by the OrderedProductsEventHandler, which will return all the existing OrderedProducts.

11. Conclusion

In this article, we introduced the Axon Framework as a powerful base for building an application leveraging the benefits of CQRS and Event Sourcing.

We implemented a simple Order service using the framework to show how such an application should be structured in practice.

Lastly, Axon Server posed as our Event Store and the message routing mechanism.

The implementation of all these examples and code snippets can be found over on GitHub.

For any additional questions you may have, also check out the Axon Framework User Group.