Backend Application with ZenWave SDK

Converting your ZDL models into a Spring Boot/Spring Cloud backend application with ZenWave SDK.

The term 'model' has a dual meaning in DDD: it's a representation of a concept in the problem space and a representation of a concept in the solution space. - Eric Evans in Domain Driven Design

With ZenWave SDK you can quickly convert your ZDL models, the results of your analysis and design, into a working Spring Boot/Spring Cloud backend application.

In this section we will guide you into the process of modeling your application into ZDL models and using ZenWave SDK plugins to generate the building blocks you can assemble to create your backend application.

Project structure

ZenWave SDK Backend Application Plugin will generate a Spring Boot/Spring Cloud backend application with the following a flexible hexagonal/onion/clean architecture:

πŸ“¦ <basePackage>
πŸ“¦ adapters
└─ web
| └─ RestControllers (spring mvc)
└─ events
└─ *EventListeners (spring-cloud-streams)
πŸ“¦ core
β”œβ”€ πŸ“¦ domain
| └─ (entities and aggregates)
β”œβ”€ πŸ“¦ inbound
| β”œβ”€ dtos/
| └─ ServiceInterface (inbound service interface)
β”œβ”€ πŸ“¦ outbound
| β”œβ”€ mongodb
| | └─ *RepositoryInterface (spring-data interface)
| └─ jpa
| └─ *RepositoryInterface (spring-data interface)
└─ πŸ“¦ implementation
β”œβ”€ mappers/
└─ ServiceImplementation (inbound service implementation)
πŸ“¦ infrastructure
β”œβ”€ mongodb
| └─ CustomRepositoryImpl (spring-data custom implementation)
└─ jpa
└─ CustomRepositoryImpl (spring-data custom implementation)

This structure can be parameterized to some extent using configuration parameters, or customized fully programmatically as explained below.

Start Modeling with ZDL

Depending on your particular use case you have different approaches to start modeling with ZDL:

  • Entities and Aggregates: Traditionally you would start your application modeling your data.
  • Services and Commands: You can also start modeling the external behaviour of your application, describing which endpoints it exposes and which commands it listens to.
  • Events: If you came from a big-picture exploratory Event-Storming you can start describing the events and the commands that trigger them.

We will start with the first approach: describing the data model, then the behaviour and last the domain events emitted to outside world, but you can use any of the other two options to start modeling your application.

Entities and Aggregates

So we start describing our data model with the entity keyword.

Adding some fields with fieldName, fieldType, validations and javadoc comments.

These javadoc comments can either precede the field definition on a separate line, typical Java convention, or be appended to the end of the line where the field is defined, which is a departure from the typical Java convention but is very handy for small comments.

For field types, you can use any valid type in java.lang, java.math.*, java.time.* or java.util.*, or reference any other entity or enum. But only these field types were tested.

Field types can also be arrays of the above types or references, but direct references and arrays only work with Mongodb, for JPA you will need to configure OneToOne, ManyToOne, ManyToOne, ManyToMany... relationships, and direct references will work only in places where JPA @Embeddable is allowed.

entity Customer {
firstName String required minlength(2) maxlength(250) /** first name javadoc comment */
lastName String required minlength(2) maxlength(250) /** last name javadoc comment */
email String required unique /** email javadoc comment */
phone String required
* address is a direct reference, which only works in mongodb (not JPA)
addresses Address[]
entity Address {
identifier String required /** Description identifier for this Address */
street String required
city String
state String
zip String
type AddressType required /** address type is an enum */
enum AddressType { HOME(1) /** home description */, WORK(1) /** work description */ }

Special annotations

ZDL Language allows any annotation name on entities and fields, it's the job of ZenWave Plugins to interpret those annotations and generate code accordingly.

The following annotations are special and will be interpreted by ZenWave SDK Backend Application Plugin in regard to entities and enums. They control aspects of generated code regarding persistence...


The most important ones are @aggregate, @embedded, @copy, @extends and @abstract.

@aggregate marks an entity as an aggregate root. Aggregate entities will have Spring Data Repository interfaces generated for them, which will be included as dependencies in you Service Implementations.

@embedded entities do not contain id and version fields and will not have a Spring Data Repository interface generated for them.

For @auditing entities it will add createdDate/By, lastModifiedDate/By fields to the entity. Usually you will use this annotation in your aggregate root entities.

With @vo it will generate a Java object without id, version and persistence annotations in the same package as persistent entities.

You can use @skip for entities you are just using to @copyproperties.

You can also reuse entity definitions or create a class hierarchy using @copy and @extends annotations, and mark some entities as @abstract.

entity BaseEntity {
someField String
entity EntityA {
// will contain all the fields from BaseEntity
entity EntityB {
// will extend BaseEntity

While @extend will create a class hierarchy, @copy will copy all the fields from the referenced entity into the current entity and can be used when you need to reuse fields in entities that belong to different packages (like domain, inputs, outputs and events)

entity BaseEntity {
someField String
input InputA {
// an input can not extend and entity because is not an accesible package

Entity fields annotated with @transient will be annotated with the appropriate @Transient persistence annotation in the generated code.

Services Commands

Services bundle together the operations that your aggregates expose to the outside world. They are the entry point to the core, innermost layer, or hexagon of your application.

In a Domain-Driven Design (DDD) sense, is the combination of entities annotated with @aggregate and the associated service commands collectively what form an Aggregate. Encapsulating a cluster of domain objects that can be treated as a transactional single unit, ensuring consistency and enforcing business rules.

This combination is sometimes referred as the Anemic Domain Model. We call it Separation of Concerns, clearly separating where data is defined and the business process you apply to it, and it works just fine for most microservice applications.

Services are still the only entry point for your aggregates functionality, and this programming model greatly simplifies how to interact with external dependencies such as persistence, APIs and publishing domain events to an external broker.

Services Section in ZDL Reference provides more details on the syntax. This section will provide you guidance on how to use them.

entity Customer {
service CustomerService for (Customer) {
/* the following service method names and signatures will genereate CRUD operations */
createCustomer(Customer) Customer withEvents CustomerEvent
updateCustomer(id, Customer) Customer? withEvents CustomerEvent
deleteCustomer(id) withEvents CustomerEvent
getCustomer(id) Customer?
listCustomers() Customer[]
/* this is will generate (almos) empty implementation placeholders for service and tests */
updateCustomerAddress(id, Address) Customer? withEvents CustomerEvent
// This will generate two methods: 'longRunningOperation' and 'longRunningOperationSync'
longRunningOperation(id) withEvents CustomerEvent

Service commands resemble but should not be confused wit java methods. They support only two types of parameters:

  • id: the presence of this parameter specifies that this command operates on an aggregate entity instance with the given id.
  • CommandInput: it can point to an entity or input type, and it's used to pass data to the command.

CommandOutput can be any valid entity or output type, or be unspecified. If CommandOutput is an entity, it expresses that entity will be created or updated with the command input data. CommandOutput can be marked as optional with a ? to express that the command may not return any output.

withEvents is used to specify the domain events that will be published after the command is executed. It can be a single event, a space separated list of events or two or more alternative events.

service CustomerService for (Customer) {
someCustomerCommand(id, Customer) Customer? withEvents CustomerEvent [AlternativeCustomerEvent|AnotherCustomerEvent]

Java signature for CommandInput can be expanded using an @inline annotation in the input type definition:

input CustomerFilter {
firstName String
lastName String
email String
service CustomerService for (Customer) {
searchCustomers(CustomerFilter) Customer[] // this will produce the following Java method signature:
// Page<Customer> searchCustomers(String firstName, String lastName, String email)

Inputs and Outputs

You can directly use entity types as command parameters, or you can create specific input and output types for service commands parameters and return types.

Using ZenWave SDK Backend Plugin option inputDTOSuffix to a non-empty string, it will automatically create an input class for your entities, copying all the fields from each entity.

Also, if you use an entity as command parameter, and you don't set inputDTOSuffix, ZenWave SDK will still use a mapstruct mapper to copy parameter fields into persistent entity. So you are always in control of how you map your input data into your persistent entities.

public Customer createCustomer(Customer input) {
log.debug("Request to save Customer: {}", input);
// Customer parameter is an entity, but it will be copied into a persistent entity here
var customer = customerServiceMapper.update(new Customer(), input);
customer =;
return customer;

If you need bigger flexibility you can just create your own input and output types.

input CustomerInput {
firstName String required minlength(2) maxlength(250)
lastName String required minlength(2) maxlength(250)
email String required unique
phone String required
output CustomerOutput {
firstName String required minlength(2) maxlength(250)
lastName String required minlength(2) maxlength(250)
email String required unique
phone String required

If you need to create an enum in the input package, you can use the @input annotation.

@input // this will create an enum in the input package
enum CustomerStatusInput { ACTIVE, INACTIVE }

Exposing your Services to Outside World

Your services represent the core of your application and are not exposed by default to the outside world, through adapters or primary ports.

ZenWave proposed architecture is to use API-First to describe your adapters and primary ports interface.

You can still annotate your Service Commands with @rest and @asyncapiannotations for two different purposes:

  1. Document how each service command will be exposed to the outside world.
  2. Generate draft versions of AsyncAPI and OpenAPI from your ZDL models. Then you can generate code from those API-First specifications using ZenWave AsyncAPI and OpenAPI Plugins.

ZDL Language is very compact and concise for this, but remember that API-First definitions are the source of truth for outside communication.

With ZDLToAsyncAPIPlugin and ZDLToOpenAPIPlugin plugins you can create complete draft versions of AsyncAPI and OpenAPI specifications from your annotated Services and Events.

* Customer Service annotated for REST and AsyncAPI serves two purposes:
* 1. Document how each service command will be exposed to the outside world.
* 2. Generate draft versions of AsyncAPI and OpenAPI from your ZDL models.
service CustomerService for (Customer) {
createCustomer(Customer) Customer withEvents CustomerEvent
updateCustomer(id, Customer) Customer? withEvents CustomerEvent
updateCustomerAddress(id, AddressInput) Customer? withEvents CustomerEvent CustomerAddressUpdated
deleteCustomer(id) withEvents CustomerEvent
getCustomer(id) Customer?
@get({params: {search: "string"}}) @paginated
listCustomers() Customer[]
@skip // skip generating this domain enum, it will genereate by asyncapi code generator.
enum EventType { CREATED(1), UPDATED(1), DELETED(1) }
@asyncapi({channel: "CustomerEventsChannel", topic: ""})
event CustomerEvent {
customerId String
eventType EventType
customer Customer
@asyncapi({channel: "CustomerAddressEventsChannel", topic: "customer.address-events"})
event CustomerAddressUpdated {
customerId String
addressDescription String
originalAddress Address
newAddress Address

See Producing Domain Events, Consuming Async Commands and Events and Exposing a REST API for specific details on how to use ZenWave SDK Plugins to generate code for your adapters and primary ports.


Events carry the information about relevant changes in your bounded context. They are meant to be published to outside world and documented through and API-First specification like AsyncAPI.

ZenWave SDK Backend Plugin can generate the code for publishing events as part of your service commands, but it does not generate code for the events data structures, event data models will be generated by ZenWave AsyncAPI Plugins.

AsyncAPI definitions are the source of truth for outside communication, still ZDL events work as a very compact and concise IDL to generate your AsyncAPI definition.

For instance the following definition will generate an AsyncAPI definition with:

@asyncapi({channel: "OrderUpdatesChannel", topic: "orders.order_updates"})
event OrderStatusUpdated {
id String
dateTime Instant required
status OrderStatus required
previousStatus OrderStatus
  • An schema named OrderStatusUpdated with a payload containing the id, dateTime, status and previousStatus fields.
  • A message named OrderStatusUpdatedMessage pointing to OrderStatusUpdated schema.
  • An a Channel named OrderUpdatesChannel containing a reference to the OrderStatusUpdatedMessage message.
  • It also will generate an Operation named onOrderStatusUpdated with and action sendto the OrderUpdatesChannel channel.

This is as a compact format as it can get!! Saving you a lot of typing and giving you very concise representation of your events.

NOTE: you still need to add this event to a serviceCommand to generate the asyncapi definition for it.

service OrdersService for (CustomerOrder) {
// only emited events will be included in the asyncapi definition
updateOrder(id, CustomerOrderInput) CustomerOrder withEvents OrderStatusUpdated

Including Event Publishing Code

Because this depends on ZenWave AsyncAPI Plugins naming conventions, you need to explicitly set includeEmitEventsImplementation to true in ZenWave SDK Backend Plugin option to include event publishing code in your core services.

// This will only be generated if includeEmitEventsImplementation is set to true
private final EventsMapper eventsMapper = EventsMapper.INSTANCE;
private final ICustomerEventsProducer eventsProducer;
public Customer createCustomer(Customer input) {
log.debug("Request to save Customer: {}", input);
var customer = customerServiceMapper.update(new Customer(), input);
customer =;
// This will only be generated if includeEmitEventsImplementation is set to true
var customerEvent = eventsMapper.asCustomerEvent(customer);
return customer;

Generating Code with CLI or IntelliJ Plugin

Now you can ZenWave SDK BackendApplicationDefaultPlugin plugin to generate your backend application.

From the command line using JBang:

jbang zw -p BackendApplicationDefaultPlugin \
specFile=orders.zdl \
basePackage="io.zenwave360.example.orders" \
persistence=mongodb \
useLombok=true \
includeEmitEventsImplementation=true \

Or using ZenWave Model Editor in IntelliJ IDEA:

ZenWave SDK Backend Generation with IntelliJ Plugins

Notes about Generated Hexagonal Architecture

ZenWave SDK Modeling Languages

ZenWave SDK Backend Application Plugin will generate a Spring Boot/Spring Cloud backend application with the following a flexible hexagonal/onion/clean architecture.

For the sake of simplicity, it makes the following two concessions:

  • Domain entities are annotated with JPA or SpringData Mongodb annotations.
  • Spring Data Repository interfaces are used as outbound adapters or secondary ports interfaces. While the implementation is outside the core domain layer.

These two concessions will simplify the number of abstractions and layers of Mappers and DTOs that you need to create (and maintain), while preserving the benefits of a hexagonal/onion/clean architecture (inside vs outside).

In fact, ZenWave SDK Backend Plugin provides in-memory implementations of Spring Data Repository interfaces, so you can test your application without the need of a database (or Mockito). Proving that you are not tied to any particular persistence implementation.

If you feel that the very unlikely case that you need to change your persistence implementation is worth the extra complexity of adding layers of mappers and DTOs and persistence delegates, you can always create a custom generator plugin like the to generate them.

Project Structure Customization

Generated project package names can be (to some extent) customized via 'undocumented' plugin properties. If you need them you can dig into the source code to find them.

We believe that microservices do not need multi-module maven projects, but if you believe otherwise you can customize this by creating a custom generator plugin using, which can generate multi-module projects, as a starting point.

ZenWave SDK Backend Application Plugin will generate an ArchitectureTest using ArchUnit to validate that your application follows hexagonal/onion/clean architecture and follow strict package import rules.