Spring Cloud Function

Mark Fisher, Dave Syer, Oleg Zhurakousky, Anshul Mehra

3.0.1.RELEASE

Introduction

Spring Cloud Function is a project with the following high-level goals:

Promote the implementation of business logic via functions.
Decouple the development lifecycle of business logic from any specific runtime target so that the same code can run as a web endpoint, a stream processor, or a task.
Support a uniform programming model across serverless providers, as well as the ability to run standalone (locally or in a PaaS).
Enable Spring Boot features (auto-configuration, dependency injection, metrics) on serverless providers.

It abstracts away all of the transport details and infrastructure, allowing the developer to keep all the familiar tools and processes, and focus firmly on business logic.

Here’s a complete, executable, testable Spring Boot application (implementing a simple string manipulation):

@SpringBootApplication
public class Application {

  @Bean
  public Function<Flux<String>, Flux<String>> uppercase() {
    return flux -> flux.map(value -> value.toUpperCase());
  }

  public static void main(String[] args) {
    SpringApplication.run(Application.class, args);
  }
}

It’s just a Spring Boot application, so it can be built, run and tested, locally and in a CI build, the same way as any other Spring Boot application. The Function is from java.util and Flux is a Reactive Streams Publisher from Project Reactor. The function can be accessed over HTTP or messaging.

Spring Cloud Function has 4 main features:

In the nutshell Spring Cloud Function provides the following features: 1. Wrappers for @Beans of type Function, Consumer and Supplier, exposing them to the outside world as either HTTP endpoints and/or message stream listeners/publishers with RabbitMQ, Kafka etc.

Choice of programming styles - reactive, imperative or hybrid.
Function composition and adaptation (e.g., composing imperative functions with reactive).
Support for reactive function with multiple inputs and outputs allowing merging, joining and other complex streaming operation to be handled by functions.
Transparent type conversion of inputs and outputs.
Packaging functions for deployments, specific to the target platform (e.g., Project Riff, AWS Lambda and more)
Adapters to expose function to the outside world as HTTP endpoints etc.
Deploying a JAR file containing such an application context with an isolated classloader, so that you can pack them together in a single JVM.
Compiling strings which are Java function bodies into bytecode, and then turning them into @Beans that can be wrapped as above.
Adapters for AWS Lambda, Azure, Apache OpenWhisk and possibly other "serverless" service providers.

Spring Cloud is released under the non-restrictive Apache 2.0 license. If you would like to contribute to this section of the documentation or if you find an error, please find the source code and issue trackers in the project at github.

Getting Started

Build from the command line (and "install" the samples):

$ ./mvnw clean install

(If you like to YOLO add -DskipTests.)

Run one of the samples, e.g.

$ java -jar spring-cloud-function-samples/function-sample/target/*.jar

This runs the app and exposes its functions over HTTP, so you can convert a string to uppercase, like this:

$ curl -H "Content-Type: text/plain" localhost:8080/uppercase -d Hello
HELLO

You can convert multiple strings (a Flux<String>) by separating them with new lines

$ curl -H "Content-Type: text/plain" localhost:8080/uppercase -d 'Hello
> World'
HELLOWORLD

(You can use ^QJ in a terminal to insert a new line in a literal string like that.)

Programming model

Function Catalog and Flexible Function Signatures

One of the main features of Spring Cloud Function is to adapt and support a range of type signatures for user-defined functions, while providing a consistent execution model. That’s why all user defined functions are transformed into a canonical representation by FunctionCatalog.

While users don’t normally have to care about the FunctionCatalog at all, it is useful to know what kind of functions are supported in user code.

It is also important to understand that Spring Cloud Function provides first class support for reactive API provided by Project Reactor allowing reactive primitives such as Mono and Flux to be used as types in user defined functions providing greater flexibility when choosing programming model for your function implementation. Reactive programming model also enables functional support for features that would be otherwise difficult to impossible to implement using imperative programming style. For more on this please read Function Arity section.

Java 8 function support

Spring Cloud Function embraces and builds on top of the 3 core functional interfaces defined by Java and available to us since Java 8.

Supplier<O>
Function<I, O>
Consumer<I>

Supplier

Supplier can be reactive - Supplier<Flux<T>> or imperative - Supplier<T>. From the invocation standpoint this should make no difference to the implementor of such Supplier. However, when used within frameworks (e.g., Spring Cloud Stream), Suppliers, especially reactive, often used to represent the source of the stream, therefore they are invoked once to get the stream (e.g., Flux) to which consumers can subscribe to. In other words such suppliers represent an equivalent of an infinite stream. However, the same reactive suppliers can also represent finite stream(s) (e.g., result set on the polled JDBC data). In those cases such reactive suppliers must be hooked up to some polling mechanism of the underlying framework.

To assist with that Spring Cloud Function provides a marker annotation org.springframework.cloud.function.context.PollableSupplier to signal that such supplier produces a finite stream and may need to be polled again. That said, it is important to understand that Spring Cloud Function itself provides no behavior for this annotation.

In addition PollableSupplier annotation exposes a splittable attribute to signal that produced stream needs to be split (see Splitter EIP)

Here is the example:

@PollableSupplier(splittable = true)
public Supplier<Flux<String>> someSupplier() {
	return () -> {
		String v1 = String.valueOf(System.nanoTime());
		String v2 = String.valueOf(System.nanoTime());
		String v3 = String.valueOf(System.nanoTime());
		return Flux.just(v1, v2, v3);
	};
}

Function

Function can also be written in imperative or reactive way, yet unlike Supplier and Consumer there are no special considerations for the implementor other then understanding that when used within frameworks such as Spring Cloud Stream and others, reactive function is invoked only once to pass a reference to the stream (Flux or Mono) and imperative is invoked once per event.

Consumer

Consumer is a little bit special because it has a void return type, which implies blocking, at least potentially. Most likely you will not need to write Consumer<Flux<?>>, but if you do need to do that, remember to subscribe to the input flux. If you declare a Consumer of a non publisher type (which is normal), it will be converted to a function that returns a publisher, so that it can be subscribed to in a controlled way.

Function Composition

Function Composition is a feature that allows one to compose several functions into one. The core support is based on function composition feature available with Function.andThen(..) support available since Java 8. However on top of it, we provide few additional features.

Declarative Function Composition

This feature allows you to provide composition instruction in a declarative way using | (pipe) or , (comma) delimiter when providing spring.cloud.function.definition property.

For example

--spring.cloud.function.definition=uppercase|reverse

Here we effectively provided a definition of a single function which itself is a composition of function uppercase and function reverse. In fact that is one of the reasons why the property name is definition and not name, since the definition of a function can be a composition of several named functions. And as mentioned you can use , instead of pipe (such as …definition=uppercase,reverse).

Composing non-Functions

Spring Cloud Function also supports composing Supplier with Consumer or Function as well as Function with Consumer. What’s important here is to understand the end product of such definitions. Composing Supplier with Function still results in Supplier while composing Supplier with Consumer will effectively render Runnable. Following the same logic composing Function with Consumer will result in Consumer.

And of course you can’t compose uncomposable such as Consumer and Function, Consumer and Supplier etc.

Function Routing

Since version 2.2 Spring Cloud Function provides routing feature allowing you to invoke a single function which acts as a router to an actual function you wish to invoke This feature is very useful in certain FAAS environments where maintaining configurations for several functions could be cumbersome or exposing more then one function is not possible.

The RoutingFunction is registered in FunctionCatalog under the name functionRouter. For simplicity and consistency you can also refer to RoutingFunction.FUNCTION_NAME constant.

This function has the following signature:

public class RoutingFunction implements Function<Object, Object> {
. . .
}

The routing instructions could be communicated in several ways;

Message Headers

If the input argument is of type Message<?>, you can communicate routing instruction by setting one of spring.cloud.function.definition or spring.cloud.function.routing-expression Message headers. For more static cases you can use spring.cloud.function.definition header which allows you to provide the name of a single function (e.g., …definition=foo) or a composition instruction (e.g., …definition=foo|bar|baz). For more dynamic cases you can use spring.cloud.function.routing-expression header which allows you to use Spring Expression Language (SpEL) and provide SpEL expression that should resolve into definition of a function (as described above).

SpEL evaluation context’s root object is the actual input argument, so in he case of Message<?> you can construct expression that has access to both payload and headers (e.g., spring.cloud.function.routing-expression=headers.function_name).

In specific execution environments/models the adapters are responsible to translate and communicate spring.cloud.function.definition and/or spring.cloud.function.routing-expression via Message header. For example, when using spring-cloud-function-web you can provide spring.cloud.function.definition as an HTTP header and the framework will propagate it as well as other HTTP headers as Message headers.

Application Properties

Routing instruction can also be communicated via spring.cloud.function.definition or spring.cloud.function.routing-expression as application properties. The rules described in the previous section apply here as well. The only difference is you provide these instructions as application properties (e.g., --spring.cloud.function.definition=foo).

When dealing with reactive inputs (e.g., Publisher), routing instructions must only be provided via Function properties. This is due to the nature of the reactive functions which are invoked only once to pass a Publisher and the rest is handled by the reactor, hence we can not access and/or rely on the routing instructions communicated via individual values (e.g., Message).

Function Arity

There are times when a stream of data needs to be categorized and organized. For example, consider a classic big-data use case of dealing with unorganized data containing, let’s say, ‘orders’ and ‘invoices’, and you want each to go into a separate data store. This is where function arity (functions with multiple inputs and outputs) support comes to play.

Let’s look at an example of such a function (full implementation details are available here),

@Bean
public Function<Flux<Integer>, Tuple2<Flux<String>, Flux<String>>> organise() {
	return flux -> ...;
}

Given that Project Reactor is a core dependency of SCF, we are using its Tuple library. Tuples give us a unique advantage by communicating to us both cardinality and type information. Both are extremely important in the context of SCSt. Cardinality lets us know how many input and output bindings need to be created and bound to the corresponding inputs and outputs of a function. Awareness of the type information ensures proper type conversion.

Also, this is where the ‘index’ part of the naming convention for binding names comes into play, since, in this function, the two output binding names are organise-out-0 and organise-out-1.

IMPORTANT: At the moment, function arity is only supported for reactive functions (Function<TupleN<Flux<?>…>, TupleN<Flux<?>…>>) centered on Complex event processing where evaluation and computation on confluence of events typically requires view into a stream of events rather than single event.

Kotlin Lambda support

We also provide support for Kotlin lambdas (since v2.0). Consider the following:

@Bean
open fun kotlinSupplier(): () -> String {
    return  { "Hello from Kotlin" }
}

@Bean
open fun kotlinFunction(): (String) -> String {
    return  { it.toUpperCase() }
}

@Bean
open fun kotlinConsumer(): (String) -> Unit {
    return  { println(it) }
}

The above represents Kotlin lambdas configured as Spring beans. The signature of each maps to a Java equivalent of Supplier, Function and Consumer, and thus supported/recognized signatures by the framework. While mechanics of Kotlin-to-Java mapping are outside of the scope of this documentation, it is important to understand that the same rules for signature transformation outlined in "Java 8 function support" section are applied here as well.

To enable Kotlin support all you need is to add spring-cloud-function-kotlin module to your classpath which contains the appropriate autoconfiguration and supporting classes.

Function Component Scan

Spring Cloud Function will scan for implementations of Function, Consumer and Supplier in a package called functions if it exists. Using this feature you can write functions that have no dependencies on Spring - not even the @Component annotation is needed. If you want to use a different package, you can set spring.cloud.function.scan.packages. You can also use spring.cloud.function.scan.enabled=false to switch off the scan completely.

Standalone Web Applications

The spring-cloud-function-web module has autoconfiguration that activates when it is included in a Spring Boot web application (with MVC support). There is also a spring-cloud-starter-function-web to collect all the optional dependencies in case you just want a simple getting started experience.

With the web configurations activated your app will have an MVC endpoint (on "/" by default, but configurable with spring.cloud.function.web.path) that can be used to access the functions in the application context. The supported content types are plain text and JSON.

Method	Path	Request	Response	Status
GET	/{supplier}	-	Items from the named supplier	200 OK
POST	/{consumer}	JSON object or text	Mirrors input and pushes request body into consumer	202 Accepted
POST	/{consumer}	JSON array or text with new lines	Mirrors input and pushes body into consumer one by one	202 Accepted
POST	/{function}	JSON object or text	The result of applying the named function	200 OK
POST	/{function}	JSON array or text with new lines	The result of applying the named function	200 OK
GET	/{function}/{item}	-	Convert the item into an object and return the result of applying the function	200 OK

Method

Path

Request

Response

Status

GET

/{supplier}

Items from the named supplier

200 OK

POST

/{consumer}

JSON object or text

Mirrors input and pushes request body into consumer

202 Accepted

POST

/{consumer}

JSON array or text with new lines

Mirrors input and pushes body into consumer one by one

202 Accepted

POST

/{function}

JSON object or text

The result of applying the named function

200 OK

POST

/{function}

JSON array or text with new lines

The result of applying the named function

200 OK

GET

/{function}/{item}

Convert the item into an object and return the result of applying the function

200 OK

As the table above shows the behaviour of the endpoint depends on the method and also the type of incoming request data. When the incoming data is single valued, and the target function is declared as obviously single valued (i.e. not returning a collection or Flux), then the response will also contain a single value. For multi-valued responses the client can ask for a server-sent event stream by sending `Accept: text/event-stream".

If there is only a single function (consumer etc.) in the catalog, the name in the path is optional. Composite functions can be addressed using pipes or commas to separate function names (pipes are legal in URL paths, but a bit awkward to type on the command line).

For cases where there is more then a single function in catalog and you want to map a specific function to the root path (e.g., "/"), or you want to compose several functions and then map to the root path you can do so by providing spring.cloud.function.definition property which essentially used by spring-cloud-function-web module to provide default mapping for cases where there is some type of a conflict (e.g., more then one function available etc).

For example,

--spring.cloud.function.definition=foo|bar

The above property will compose 'foo' and 'bar' function and map the composed function to the "/" path.

Functions and consumers that are declared with input and output in Message<?> will see the request headers on the input messages, and the output message headers will be converted to HTTP headers.

When POSTing text the response format might be different with Spring Boot 2.0 and older versions, depending on the content negotiation (provide content type and accpt headers for the best results).

See Testing Functional Applications to see the details and example on how to test such application.

Standalone Streaming Applications

To send or receive messages from a broker (such as RabbitMQ or Kafka) you can leverage spring-cloud-stream project and it’s integration with Spring Cloud Function. Please refer to Spring Cloud Function section of the Spring Cloud Stream reference manual for more details and examples.

Deploying a Packaged Function

Spring Cloud Function provides a "deployer" library that allows you to launch a jar file (or exploded archive, or set of jar files) with an isolated class loader and expose the functions defined in it. This is quite a powerful tool that would allow you to, for instance, adapt a function to a range of different input-output adapters without changing the target jar file. Serverless platforms often have this kind of feature built in, so you could see it as a building block for a function invoker in such a platform (indeed the Riff Java function invoker uses this library).

The standard entry point is to add spring-cloud-function-deployer to the classpath, the deployer kicks in and looks for some configuration to tell it where to find the function jar.

<dependency>
	<groupId>org.springframework.cloud</groupId>
	<artifactId>spring-cloud-function-deployer</artifactId>
	<version>${spring.cloud.function.version}</version>
</dependency>

At a minimum the user has to provide a spring.cloud.function.location which is a URL or resource location for the archive containing the functions. It can optionally use a maven: prefix to locate the artifact via a dependency lookup (see FunctionProperties for complete details). A Spring Boot application is bootstrapped from the jar file, using the MANIFEST.MF to locate a start class, so that a standard Spring Boot fat jar works well, for example. If the target jar can be launched successfully then the result is a function registered in the main application’s FunctionCatalog. The registered function can be applied by code in the main application, even though it was created in an isolated class loader (by deault).

Here is the example of deploying a JAR which contains an 'uppercase' function and invoking it .

@SpringBootApplication
public class DeployFunctionDemo {

	public static void main(String[] args) {
		ApplicationContext context = SpringApplication.run(DeployFunctionDemo.class,
				"--spring.cloud.function.location=..../target/uppercase-0.0.1-SNAPSHOT.jar",
				"--spring.cloud.function.definition=uppercase");

		FunctionCatalog catalog = context.getBean(FunctionCatalog.class);
		Function<String, String> function = catalog.lookup("uppercase");
		System.out.println(function.apply("hello"));
	}
}

Functional Bean Definitions

Spring Cloud Function supports a "functional" style of bean declarations for small apps where you need fast startup. The functional style of bean declaration was a feature of Spring Framework 5.0 with significant enhancements in 5.1.

Comparing Functional with Traditional Bean Definitions

Here’s a vanilla Spring Cloud Function application from with the familiar @Configuration and @Bean declaration style:

@SpringBootApplication
public class DemoApplication {

  @Bean
  public Function<String, String> uppercase() {
    return value -> value.toUpperCase();
  }

  public static void main(String[] args) {
    SpringApplication.run(DemoApplication.class, args);
  }

}

Now for the functional beans: the user application code can be recast into "functional" form, like this:

@SpringBootConfiguration
public class DemoApplication implements ApplicationContextInitializer<GenericApplicationContext> {

  public static void main(String[] args) {
    FunctionalSpringApplication.run(DemoApplication.class, args);
  }

  public Function<String, String> uppercase() {
    return value -> value.toUpperCase();
  }

  @Override
  public void initialize(GenericApplicationContext context) {
    context.registerBean("demo", FunctionRegistration.class,
        () -> new FunctionRegistration<>(uppercase())
            .type(FunctionType.from(String.class).to(String.class)));
  }

}

The main differences are:

The main class is an ApplicationContextInitializer.
The @Bean methods have been converted to calls to context.registerBean()
The @SpringBootApplication has been replaced with @SpringBootConfiguration to signify that we are not enabling Spring Boot autoconfiguration, and yet still marking the class as an "entry point".
The SpringApplication from Spring Boot has been replaced with a FunctionalSpringApplication from Spring Cloud Function (it’s a subclass).

The business logic beans that you register in a Spring Cloud Function app are of type FunctionRegistration. This is a wrapper that contains both the function and information about the input and output types. In the @Bean form of the application that information can be derived reflectively, but in a functional bean registration some of it is lost unless we use a FunctionRegistration.

An alternative to using an ApplicationContextInitializer and FunctionRegistration is to make the application itself implement Function (or Consumer or Supplier). Example (equivalent to the above):

@SpringBootConfiguration
public class DemoApplication implements Function<String, String> {

  public static void main(String[] args) {
    FunctionalSpringApplication.run(DemoApplication.class, args);
  }

  @Override
  public String uppercase(String value) {
    return value.toUpperCase();
  }

}

It would also work if you add a separate, standalone class of type Function and register it with the SpringApplication using an alternative form of the run() method. The main thing is that the generic type information is available at runtime through the class declaration.

Suppose you have

@Component
public class CustomFunction implements Function<Flux<Foo>, Flux<Bar>> {
	@Override
	public Flux<Bar> apply(Flux<Foo> flux) {
		return flux.map(foo -> new Bar("This is a Bar object from Foo value: " + foo.getValue()));
	}

}

You register it as such:

@Override
public void initialize(GenericApplicationContext context) {
		context.registerBean("function", FunctionRegistration.class,
				() -> new FunctionRegistration<>(new CustomFunction()).type(CustomFunction.class));
}

Limitations of Functional Bean Declaration

Most Spring Cloud Function apps have a relatively small scope compared to the whole of Spring Boot, so we are able to adapt it to these functional bean definitions easily. If you step outside that limited scope, you can extend your Spring Cloud Function app by switching back to @Bean style configuration, or by using a hybrid approach. If you want to take advantage of Spring Boot autoconfiguration for integrations with external datastores, for example, you will need to use @EnableAutoConfiguration. Your functions can still be defined using the functional declarations if you want (i.e. the "hybrid" style), but in that case you will need to explicitly switch off the "full functional mode" using spring.functional.enabled=false so that Spring Boot can take back control.

Testing Functional Applications

Spring Cloud Function also has some utilities for integration testing that will be very familiar to Spring Boot users.

Suppose this is your application:

@SpringBootApplication
public class SampleFunctionApplication {

    public static void main(String[] args) {
        SpringApplication.run(SampleFunctionApplication.class, args);
    }

    @Bean
    public Function<String, String> uppercase() {
        return v -> v.toUpperCase();
    }
}

Here is an integration test for the HTTP server wrapping this application:

@SpringBootTest(classes = SampleFunctionApplication.class,
            webEnvironment = WebEnvironment.RANDOM_PORT)
public class WebFunctionTests {

    @Autowired
    private TestRestTemplate rest;

    @Test
    public void test() throws Exception {
        ResponseEntity<String> result = this.rest.exchange(
            RequestEntity.post(new URI("/uppercase")).body("hello"), String.class);
        System.out.println(result.getBody());
    }
}

or when function bean definition style is used:

@FunctionalSpringBootTest
public class WebFunctionTests {

    @Autowired
    private TestRestTemplate rest;

    @Test
    public void test() throws Exception {
        ResponseEntity<String> result = this.rest.exchange(
            RequestEntity.post(new URI("/uppercase")).body("hello"), String.class);
        System.out.println(result.getBody());
    }
}

This test is almost identical to the one you would write for the @Bean version of the same app - the only difference is the @FunctionalSpringBootTest annotation, instead of the regular @SpringBootTest. All the other pieces, like the @Autowired TestRestTemplate, are standard Spring Boot features.

And to help with correct dependencies here is the excerpt from POM

    <parent>
        <groupId>org.springframework.boot</groupId>
        <artifactId>spring-boot-starter-parent</artifactId>
        <version>2.2.2.RELEASE</version>
        <relativePath/> <!-- lookup parent from repository -->
    </parent>
    . . . .
    <dependency>
        <groupId>org.springframework.cloud</groupId>
        <artifactId>spring-cloud-function-web</artifactId>
        <version>3.0.1.BUILD-SNAPSHOT</version>
    </dependency>
    <dependency>
        <groupId>org.springframework.boot</groupId>
        <artifactId>spring-boot-starter</artifactId>
    </dependency>
    <dependency>
        <groupId>org.springframework.boot</groupId>
        <artifactId>spring-boot-starter-web</artifactId>
        <scope>test</scope>
    </dependency>
    <dependency>
        <groupId>org.springframework.boot</groupId>
        <artifactId>spring-boot-starter-test</artifactId>
        <scope>test</scope>
        <exclusions>
            <exclusion>
                <groupId>org.junit.vintage</groupId>
                <artifactId>junit-vintage-engine</artifactId>
            </exclusion>
        </exclusions>
    </dependency>

Or you could write a test for a non-HTTP app using just the FunctionCatalog. For example:

@RunWith(SpringRunner.class)
@FunctionalSpringBootTest
public class FunctionalTests {

	@Autowired
	private FunctionCatalog catalog;

	@Test
	public void words() throws Exception {
		Function<String, String> function = catalog.lookup(Function.class,
				"uppercase");
		assertThat(function.apply("hello")).isEqualTo("HELLO");
	}

}

Dynamic Compilation

There is a sample app that uses the function compiler to create a function from a configuration property. The vanilla "function-sample" also has that feature. And there are some scripts that you can run to see the compilation happening at run time. To run these examples, change into the scripts directory:

cd scripts

Also, start a RabbitMQ server locally (e.g. execute rabbitmq-server).

Start the Function Registry Service:

./function-registry.sh

./registerFunction.sh -n uppercase -f "f->f.map(s->s.toString().toUpperCase())"

Run a REST Microservice using that Function:

./web.sh -f uppercase -p 9000
curl -H "Content-Type: text/plain" -H "Accept: text/plain" localhost:9000/uppercase -d foo

./registerSupplier.sh -n words -f "()->Flux.just(\"foo\",\"bar\")"

Run a REST Microservice using that Supplier:

./web.sh -s words -p 9001
curl -H "Accept: application/json" localhost:9001/words

./registerConsumer.sh -n print -t String -f "System.out::println"

Run a REST Microservice using that Consumer:

./web.sh -c print -p 9002
curl -X POST -H "Content-Type: text/plain" -d foo localhost:9002/print

Run Stream Processing Microservices:

First register a streaming words supplier:

./registerSupplier.sh -n wordstream -f "()->Flux.interval(Duration.ofMillis(1000)).map(i->\"message-\"+i)"

Then start the source (supplier), processor (function), and sink (consumer) apps (in reverse order):

./stream.sh -p 9103 -i uppercaseWords -c print
./stream.sh -p 9102 -i words -f uppercase -o uppercaseWords
./stream.sh -p 9101 -s wordstream -o words

The output will appear in the console of the sink app (one message per second, converted to uppercase):

MESSAGE-0
MESSAGE-1
MESSAGE-2
MESSAGE-3
MESSAGE-4
MESSAGE-5
MESSAGE-6
MESSAGE-7
MESSAGE-8
MESSAGE-9
...

Serverless Platform Adapters

As well as being able to run as a standalone process, a Spring Cloud Function application can be adapted to run one of the existing serverless platforms. In the project there are adapters for AWS Lambda, Azure, and Apache OpenWhisk. The Oracle Fn platform has its own Spring Cloud Function adapter. And Riff supports Java functions and its Java Function Invoker acts natively is an adapter for Spring Cloud Function jars.

AWS Lambda

The AWS adapter takes a Spring Cloud Function app and converts it to a form that can run in AWS Lambda.

The details of how to get stared with AWS Lambda is out of scope of this document, so the expectation is that user has some familiarity with AWS and AWS Lambda and wants to learn what additional value spring provides.

Getting Started

One of the goals of Spring Cloud Function framework is to provide necessary infrastructure elements to enable a simple function application to interact in a certain way in a particular environment. A simple function application (in context or Spring) is an application that contains beans of type Supplier, Function or Consumer. So, with AWS it means that a simple function bean should somehow be recognised and executed in AWS Lambda environment.

Let’s look at the example:

@SpringBootApplication
public class FunctionConfiguration {

	public static void main(String[] args) {
		SpringApplication.run(FunctionConfiguration.class, args);
	}

	@Bean
	public Function<String, String> uppercase() {
		return value -> value.toUpperCase();
	}
}

It shows a complete Spring Boot application with a function bean defined in it. What’s interesting is that on the surface this is just another boot app, but in the context of AWS Adapter it is also a perfectly valid AWS Lambda application. No other code or configuration is required. All you need to do is package it and deploy it, so let’s look how we can do that.

To make things simpler we’ve provided a sample project ready to be built and deployed and you can access it here.

You simply execute ./mvnw clean package to generate JAR file. All the necessary maven plugins have already been setup to generate appropriate AWS deployable JAR file. (You can read more details about JAR layout in Notes on JAR Layout).

Then you have to upload the JAR file (via AWS dashboard or AWS CLI) to AWS.

When ask about handler you specify org.springframework.cloud.function.adapter.aws.FunctionInvoker::handleRequest which is a generic request handler.

That is all. Save and execute the function with some sample data which for this function is expected to be a String which function will uppercase and return back.

While org.springframework.cloud.function.adapter.aws.FunctionInvoker is a general purpose AWS’s RequestHandler implementation aimed at completely isolating you from the specifics of AWS Lambda API, for some cases you may want to specify which specific AWS’s RequestHandler you want to use. The next section will explain you how you can accomplish just that.

AWS Request Handlers

The adapter has a couple of generic request handlers that you can use. The most generic is (and the one we used in the Getting Started section) is org.springframework.cloud.function.adapter.aws.FunctionInvoke which is the implementation of AWS’s RequestStreamHandler. User doesn’t need to do anything other then specify it as 'handler' on AWS dashborad when deplioyimng function. It will handle most of the case including Kinesis, streaming etc. .

The most generic is SpringBootStreamHandler, which uses a Jackson ObjectMapper provided by Spring Boot to serialize and deserialize the objects in the function. There is also a SpringBootRequestHandler which you can extend, and provide the input and output types as type parameters (enabling AWS to inspect the class and do the JSON conversions itself).

If your app has more than one @Bean of type Function etc. then you can choose the one to use by configuring function.name (e.g. as FUNCTION_NAME environment variable in AWS). The functions are extracted from the Spring Cloud FunctionCatalog (searching first for Function then Consumer and finally Supplier).

Notes on JAR Layout

You don’t need the Spring Cloud Function Web or Stream adapter at runtime in Lambda, so you might need to exclude those before you create the JAR you send to AWS. A Lambda application has to be shaded, but a Spring Boot standalone application does not, so you can run the same app using 2 separate jars (as per the sample). The sample app creates 2 jar files, one with an aws classifier for deploying in Lambda, and one executable (thin) jar that includes spring-cloud-function-web at runtime. Spring Cloud Function will try and locate a "main class" for you from the JAR file manifest, using the Start-Class attribute (which will be added for you by the Spring Boot tooling if you use the starter parent). If there is no Start-Class in your manifest you can use an environment variable or system property MAIN_CLASS when you deploy the function to AWS.

If you are not using the functional bean definitions but relying on Spring Boot’s auto-configuration, then additional transformers must be configured as part of the maven-shade-plugin execution.

<plugin>
	<groupId>org.apache.maven.plugins</groupId>
	<artifactId>maven-shade-plugin</artifactId>
	<dependencies>
		<dependency>
			<groupId>org.springframework.boot</groupId>
			<artifactId>spring-boot-maven-plugin</artifactId>
		</dependency>
	</dependencies>
	<configuration>
		<createDependencyReducedPom>false</createDependencyReducedPom>
		<shadedArtifactAttached>true</shadedArtifactAttached>
		<shadedClassifierName>aws</shadedClassifierName>
		<transformers>
			<transformer implementation="org.apache.maven.plugins.shade.resource.AppendingTransformer">
				<resource>META-INF/spring.handlers</resource>
			</transformer>
			<transformer implementation="org.springframework.boot.maven.PropertiesMergingResourceTransformer">
				<resource>META-INF/spring.factories</resource>
			</transformer>
			<transformer implementation="org.apache.maven.plugins.shade.resource.AppendingTransformer">
				<resource>META-INF/spring.schemas</resource>
			</transformer>
		</transformers>
	</configuration>
</plugin>

Build file setup

In order to run Spring Cloud Function applications on AWS Lambda, you can leverage Maven or Gradle plugins offered by the cloud platform provider.

Maven

In order to use the adapter plugin for Maven, add the plugin dependency to your pom.xml file:

<dependencies>
	<dependency>
		<groupId>org.springframework.cloud</groupId>
		<artifactId>spring-cloud-function-adapter-aws</artifactId>
	</dependency>
</dependencies>

As pointed out in the Notes on JAR Layout, you wil need a shaded jar in order to upload it to AWS Lambda. You can use the Maven Shade Plugin for that. The example of the setup can be found above.

You can use theSpring Boot Maven Plugin to generate the thin jar.

<plugin>
	<groupId>org.springframework.boot</groupId>
	<artifactId>spring-boot-maven-plugin</artifactId>
	<dependencies>
		<dependency>
			<groupId>org.springframework.boot.experimental</groupId>
			<artifactId>spring-boot-thin-layout</artifactId>
			<version>${wrapper.version}</version>
		</dependency>
	</dependencies>
</plugin>

You can find the entire sample pom.xml file for deploying Spring Cloud Function applications to AWS Lambda with Maven here.

Gradle

In order to use the adapter plugin for Gradle, add the dependency to your build.gradle file:

dependencies {
	compile("org.springframework.cloud:spring-cloud-function-adapter-aws:${version}")
}

As pointed out in Notes on JAR Layout, you wil need a shaded jar in order to upload it to AWS Lambda. You can use the Gradle Shadow Plugin for that:

buildscript {
	dependencies {
		classpath "com.github.jengelman.gradle.plugins:shadow:${shadowPluginVersion}"
	}
}
apply plugin: 'com.github.johnrengelman.shadow'

assemble.dependsOn = [shadowJar]

import com.github.jengelman.gradle.plugins.shadow.transformers.*

shadowJar {
	classifier = 'aws'
	dependencies {
		exclude(
			dependency("org.springframework.cloud:spring-cloud-function-web:${springCloudFunctionVersion}"))
	}
	// Required for Spring
	mergeServiceFiles()
	append 'META-INF/spring.handlers'
	append 'META-INF/spring.schemas'
	append 'META-INF/spring.tooling'
	transform(PropertiesFileTransformer) {
		paths = ['META-INF/spring.factories']
		mergeStrategy = "append"
	}
}

You can use the Spring Boot Gradle Plugin and Spring Boot Thin Gradle Plugin to generate the thin jar.

buildscript {
	dependencies {
		classpath("org.springframework.boot.experimental:spring-boot-thin-gradle-plugin:${wrapperVersion}")
		classpath("org.springframework.boot:spring-boot-gradle-plugin:${springBootVersion}")
	}
}
apply plugin: 'org.springframework.boot'
apply plugin: 'org.springframework.boot.experimental.thin-launcher'
assemble.dependsOn = [thinJar]

You can find the entire sample build.gradle file for deploying Spring Cloud Function applications to AWS Lambda with Gradle here.

Upload

Build the sample under spring-cloud-function-samples/function-sample-aws and upload the -aws jar file to Lambda. The handler can be example.Handler or org.springframework.cloud.function.adapter.aws.SpringBootStreamHandler (FQN of the class, not a method reference, although Lambda does accept method references).

./mvnw -U clean package

Using the AWS command line tools it looks like this:

aws lambda create-function --function-name Uppercase --role arn:aws:iam::[USERID]:role/service-role/[ROLE] --zip-file fileb://function-sample-aws/target/function-sample-aws-2.0.0.BUILD-SNAPSHOT-aws.jar --handler org.springframework.cloud.function.adapter.aws.SpringBootStreamHandler --description "Spring Cloud Function Adapter Example" --runtime java8 --region us-east-1 --timeout 30 --memory-size 1024 --publish

The input type for the function in the AWS sample is a Foo with a single property called "value". So you would need this to test it:

{
  "value": "test"
}

The AWS sample app is written in the "functional" style (as an ApplicationContextInitializer). This is much faster on startup in Lambda than the traditional @Bean style, so if you don’t need @Beans (or @EnableAutoConfiguration) it’s a good choice. Warm starts are not affected.

Type Conversion

Spring Cloud Function will attempt to transparently handle type conversion between the raw input stream and types declared by your function.

For example, if your function signature is as such Function<Foo, Bar> we will attempt to convert incoming stream event to an instance of Foo.

In the event type is not known or can not be determined (e.g., Function<?, ?>) we will attempt to convert an incoming stream event to a generic Map.

Raw Input

There are times when you may want to have access to a raw input. In this case all you need is to declare your function signature to accept InputStream. For example, Function<InputStream, ?>. In this case we will not attempt any conversion and will pass the raw input directly to a function.

Microsoft Azure

The Azure adapter bootstraps a Spring Cloud Function context and channels function calls from the Azure framework into the user functions, using Spring Boot configuration where necessary. Azure Functions has quite a unique, but invasive programming model, involving annotations in user code that are specific to the platform. The easiest way to use it with Spring Cloud is to extend a base class and write a method in it with the @FunctionName annotation which delegates to a base class method.

This project provides an adapter layer for a Spring Cloud Function application onto Azure. You can write an app with a single @Bean of type Function and it will be deployable in Azure if you get the JAR file laid out right.

There is an AzureSpringBootRequestHandler which you must extend, and provide the input and output types as annotated method parameters (enabling Azure to inspect the class and create JSON bindings). The base class has two useful methods (handleRequest and handleOutput) to which you can delegate the actual function call, so mostly the function will only ever have one line.

Example:

public class FooHandler extends AzureSpringBootRequestHandler<Foo, Bar> {
	@FunctionName("uppercase")
	public Bar execute(@HttpTrigger(name = "req", methods = {HttpMethod.GET,
			HttpMethod.POST}, authLevel = AuthorizationLevel.ANONYMOUS) HttpRequestMessage<Optional<Foo>> request,
		ExecutionContext context) {
		return handleRequest(request.getBody().get(), context);
	}
}

This Azure handler will delegate to a Function<Foo,Bar> bean (or a Function<Publisher<Foo>,Publisher<Bar>>). Some Azure triggers (e.g. @CosmosDBTrigger) result in a input type of List and in that case you can bind to List in the Azure handler, or String (the raw JSON). The List input delegates to a Function with input type Map<String,Object>, or Publisher or List of the same type. The output of the Function can be a List (one-for-one) or a single value (aggregation), and the output binding in the Azure declaration should match.

If your app has more than one @Bean of type Function etc. then you can choose the one to use by configuring function.name. Or if you make the @FunctionName in the Azure handler method match the function name it should work that way (also for function apps with multiple functions). The functions are extracted from the Spring Cloud FunctionCatalog so the default function names are the same as the bean names.

Accessing Azure ExecutionContext

Some time there is a need to access the target execution context provided by Azure runtime in the form of com.microsoft.azure.functions.ExecutionContext. For example one of such needs is logging, so it can appear in the Azure console.

For that purpose Spring Cloud Function will register ExecutionContext as bean in the Application context, so it could be injected into your function. For example

@Bean
public Function<Foo, Bar> uppercase(ExecutionContext targetContext) {
	return foo -> {
		targetContext.getLogger().info("Invoking 'uppercase' on " + foo.getValue());
		return new Bar(foo.getValue().toUpperCase());
	};
}

Normally type-based injection should suffice, however if need to you can also utilise the bean name under which it is registered which is targetExecutionContext.

Notes on JAR Layout

You don’t need the Spring Cloud Function Web at runtime in Azure, so you can exclude this before you create the JAR you deploy to Azure, but it won’t be used if you include it, so it doesn’t hurt to leave it in. A function application on Azure is an archive generated by the Maven plugin. The function lives in the JAR file generated by this project. The sample creates it as an executable jar, using the thin layout, so that Azure can find the handler classes. If you prefer you can just use a regular flat JAR file. The dependencies should not be included.

Build file setup

In order to run Spring Cloud Function applications on Microsoft Azure, you can leverage the Maven plugin offered by the cloud platform provider.

In order to use the adapter plugin for Maven, add the plugin dependency to your pom.xml file:

<dependencies>
	<dependency>
		<groupId>org.springframework.cloud</groupId>
		<artifactId>spring-cloud-function-adapter-azure</artifactId>
	</dependency>
</dependencies>

Then, configure the plugin. You will need to provide Azure-specific configuration for your application, specifying the resourceGroup, appName and other optional properties, and add the package goal execution so that the function.json file required by Azure is generated for you. Full plugin documentation can be found in the plugin repository.

<plugin>
	<groupId>com.microsoft.azure</groupId>
	<artifactId>azure-functions-maven-plugin</artifactId>
	<configuration>
		<resourceGroup>${functionResourceGroup}</resourceGroup>
		<appName>${functionAppName}</appName>
	</configuration>
	<executions>
		<execution>
			<id>package-functions</id>
			<goals>
				<goal>package</goal>
			</goals>
		</execution>
	</executions>
</plugin>

You will also have to ensure that the files to be scanned by the plugin can be found in the Azure functions staging directory (see the plugin repository for more details on the staging directory and it’s default location).

You can find the entire sample pom.xml file for deploying Spring Cloud Function applications to Microsoft Azure with Maven here.

As of yet, only Maven plugin is available. Gradle plugin has not been created by the cloud platform provider.

Build

./mvnw -U clean package

Running the sample

You can run the sample locally, just like the other Spring Cloud Function samples:

and curl -H "Content-Type: text/plain" localhost:8080/api/uppercase -d '{"value": "hello foobar"}'.

You will need the az CLI app (see https://docs.microsoft.com/en-us/azure/azure-functions/functions-create-first-java-maven for more detail). To deploy the function on Azure runtime:

$ az login
$ mvn azure-functions:deploy

On another terminal try this: curl https://<azure-function-url-from-the-log>/api/uppercase -d '{"value": "hello foobar!"}'. Please ensure that you use the right URL for the function above. Alternatively you can test the function in the Azure Dashboard UI (click on the function name, go to the right hand side and click "Test" and to the bottom right, "Run").

The input type for the function in the Azure sample is a Foo with a single property called "value". So you need this to test it with something like below:

{
  "value": "foobar"
}

The Azure sample app is written in the "non-functional" style (using @Bean). The functional style (with just Function or ApplicationContextInitializer) is much faster on startup in Azure than the traditional @Bean style, so if you don’t need @Beans (or @EnableAutoConfiguration) it’s a good choice. Warm starts are not affected.