Spring Tips: Spring and GraalVM (pt. 2)

Engineering | Josh Long | June 16, 2020 | ...

speaker: Josh Long (@starbuxman)

Hi Spring fans! Welcome to a very special, interregnum episode of Spring Tips where we revisit Spring and GraalVM native images. I wanted to get this video out in light of the recent Spring Graal 0.7.1 release which drastically simplifies things even compared to the last time we looked at Spring and Graal waaay back in April 2020.

TL;DR: GraalVm is a JIT replacement that you can use with a stock-standard JVM, and that's worth looking into in its own right. GraalVM also offers a separate feature supporting native image compilation. This native-image builder takes bytecode and turns it into an architecture-specific binary that sheds the JVM and embeds something called SubstrateVM. Native images are fast to startup and they take way less memory at runtime. These qualities make it desirable in a containerized, cloud-centric environment.

In the April installment, I had to write out heaps of hand-crafted artisanal configuration. In this latest edition, it's possible to get a huge swath of applications working without variable configuration. In the video, I demonstrated how to get a Spring Data JPA (with Hibernate) and Apache Tomcat working. I also demonstrated how to get a reactive application working. Let's look at the reactive application first and then we'll look at the JPA example. The steps we'll take in the first example are common to most applications.

We're going to use GraalVM and Java 8 for this project. I'm using SDKManager to install various versions of Java: sdk install java 20.1.0.r8-grl. Then, you can optionally make it the default: sdk default java 20.1.0.r8-grl. You'll also need to install the native image builder into your GraalVM installation. Use gu install native-image. Now we can build the application.

A Reactive Example

First, go to the Spring Initializr and generate a new project with R2DBC, Lombok, H2, Reactive Web and use Java 8.

You've seen the Java code before:

package com.example.reactive;

import lombok.AllArgsConstructor;
import lombok.Data;
import lombok.NoArgsConstructor;
import lombok.RequiredArgsConstructor;
import org.springframework.boot.SpringApplication;
import org.springframework.boot.autoconfigure.SpringBootApplication;
import org.springframework.boot.autoconfigure.data.web.SpringDataWebAutoConfiguration;
import org.springframework.boot.context.event.ApplicationReadyEvent;
import org.springframework.context.ApplicationListener;
import org.springframework.data.annotation.Id;
import org.springframework.data.r2dbc.core.DatabaseClient;
import org.springframework.data.repository.reactive.ReactiveCrudRepository;
import org.springframework.stereotype.Component;
import org.springframework.web.bind.annotation.GetMapping;
import org.springframework.web.bind.annotation.RestController;
import reactor.core.publisher.Flux;

@SpringBootApplication(
        exclude = SpringDataWebAutoConfiguration.class,
        proxyBeanMethods = false
)
public class ReactiveApplication {

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

}


@RestController
@RequiredArgsConstructor
class CustomerRestController {

    private final CustomerRepository customerRepository;

    @GetMapping("/customers")
    Flux<Customer> customers() {
        return this.customerRepository.findAll();
    }
}

@Component
@RequiredArgsConstructor
class Initializer implements ApplicationListener<ApplicationReadyEvent> {

    private final CustomerRepository customerRepository;

    private final DatabaseClient databaseClient;

    @Override
    public void onApplicationEvent(ApplicationReadyEvent applicationReadyEvent) {
        Flux<Customer> save = Flux.just("Madhura", "Dr. Syer")
                .map(name -> new Customer(null, name))
                .flatMap(this.customerRepository::save);

        String sql = "create table CUSTOMER(id serial primary key, name varchar(255))";

        this.databaseClient
                .execute(sql)
                .fetch()
                .rowsUpdated()
                .thenMany(save)
                .thenMany(this.customerRepository.findAll())
                .subscribe(System.out::println);
    }
}

interface CustomerRepository extends ReactiveCrudRepository<Customer, Integer> {

}

@Data
@AllArgsConstructor
@NoArgsConstructor
class Customer {

    @Id
    private Integer id;
    private String name;

}

The only thing worth noting, as it pertains to Graal and native images anyway, is that we've disabled the creation of proxies for @Configuration class (using proxyBeanMethods = false) and excluded the SpringDataWebAutoConfiguration.class Java autoconfiguration. Hopefully, that last bit will be irrelevant in the near-term future.

That's the application. Start it and you'll see it works. We need to change the build a smidge to accommodate Graal as well. You'll need the snapshot and milestone Spring artifact repositories in your build.

    <repositories>
        <repository>
            <id>spring-milestones</id>
            <name>Spring Milestones</name>
            <url>https://repo.spring.io/milestone</url>
        </repository>
        <repository>
            <id>spring-snapshots</id>
            <name>Spring Snapshots</name>
            <url>https://repo.spring.io/snapshot</url>
            <snapshots>
                <enabled>true</enabled>
            </snapshots>
        </repository>
    </repositories>
    <pluginRepositories>
        <pluginRepository>
            <id>spring-milestones</id>
            <name>Spring Milestones</name>
            <url>https://repo.spring.io/milestone</url>
        </pluginRepository>
        <pluginRepository>
            <id>spring-snapshots</id>
            <name>Spring Snapshots</name>
            <url>https://repo.spring.io/snapshot</url>
            <snapshots>
                <enabled>true</enabled>
            </snapshots>
        </pluginRepository>
    </pluginRepositories>

Then, add these three Maven dependencies.

    <dependency>
        <groupId>org.springframework.experimental</groupId>
        <artifactId>spring-graalvm-native</artifactId>
        <version>0.7.1</version>
    </dependency>
    <dependency>
        <groupId>org.springframework.boot</groupId>
        <artifactId>spring-boot-configuration-processor</artifactId>
        <optional>true</optional>
    </dependency>
    <dependency>
        <groupId>org.springframework</groupId>
        <artifactId>spring-context-indexer</artifactId>
    </dependency>

That's it! In the code for this repository, I've also got a Spring Data MongoDB demonstration as well. It is a trivial Spring Data MongoDB application that uses Spring MVC. This example requires the exact same dependencies and attributes as the reactive one I've just shown you.

Now we'll need to compile it. First, run the normal mvn clean package. Then, we'll need to pass the .jar into the Graal native-image builder. I have a script, compile.sh, that I reuse for all three examples. Here it is.

#!/usr/bin/env bash

ARTIFACT=${1}
MAINCLASS=${2}
VERSION=${3}

JAR="${ARTIFACT}-${VERSION}.jar"

rm -rf target
mkdir -p target/native-image
mvn -ntp package  
rm -f $ARTIFACT
cd target/native-image
jar -xvf ../$JAR  
cp -R META-INF BOOT-INF/classes

LIBPATH=`find BOOT-INF/lib | tr '\n' ':'`
CP=BOOT-INF/classes:$LIBPATH
GRAALVM_VERSION=`native-image --version`

time native-image \
  --verbose \
  -H:EnableURLProtocols=http \
  -H:+RemoveSaturatedTypeFlows \
  -H:Name=$ARTIFACT \
  -Dspring.native.verbose=true \
  -Dspring.native.remove-jmx-support=true \
  -Dspring.native.remove-spel-support=true \
  -Dspring.native.remove-yaml-support=true \
  -cp $CP $MAINCLASS  

When you use this script, you need to provide three things: the build artifact, the main-class name, and the version. So, for this application, we can run it like this in the same directory:

./compile.sh reactive com.example.reactive.ReactiveApplication 0.0.1-SNAPSHOT

And then go make a cup of coffee. A quick one. Because this will take at least three minutes.

JPA

Done? Good. Let's build another example, this time using Spring Data JPA (Hibernate) and Spring MVC (with Apache Tomcat).

Go to the Spring Initializr, generate another project. This time, specify JPA, H2, Web and then click Generate. Here's the code.

package com.example.jpa;

import lombok.AllArgsConstructor;
import lombok.Data;
import lombok.NoArgsConstructor;
import lombok.RequiredArgsConstructor;
import org.springframework.boot.SpringApplication;
import org.springframework.boot.autoconfigure.SpringBootApplication;
import org.springframework.boot.autoconfigure.data.web.SpringDataWebAutoConfiguration;
import org.springframework.boot.context.event.ApplicationReadyEvent;
import org.springframework.context.ApplicationListener;
import org.springframework.context.annotation.Bean;
import org.springframework.data.jpa.repository.JpaRepository;
import org.springframework.stereotype.Component;
import org.springframework.web.bind.annotation.GetMapping;
import org.springframework.web.bind.annotation.RestController;

import javax.persistence.Entity;
import javax.persistence.GeneratedValue;
import javax.persistence.Id;
import java.util.Collection;
import java.util.stream.Stream;

@SpringBootApplication(
        exclude = SpringDataWebAutoConfiguration.class,
        proxyBeanMethods = false
)
public class JpaApplication {

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

}


@RestController
@RequiredArgsConstructor
class CustomerRestController {

    private final CustomerRepository customerRepository;

    @GetMapping("/customers")
    Collection<Customer> customers() {
        return this.customerRepository.findAll();
    }
}

@Component
@RequiredArgsConstructor
class Initializer implements ApplicationListener<ApplicationReadyEvent> {

    private final CustomerRepository customerRepository;

    @Override
    public void onApplicationEvent(ApplicationReadyEvent applicationReadyEvent) {
        Stream.of("Madhura", "Dr. Syer")
                .map(name -> new Customer(null, name))
                .map(this.customerRepository::save)
                .forEach(System.out::println);
    }
}

interface CustomerRepository extends JpaRepository<Customer, Integer> {

}

@Entity
@Data
@AllArgsConstructor
@NoArgsConstructor
class Customer {

    @Id
    @GeneratedValue
    private Integer id;
    private String name;

}

Now, this application uses JPA (and Hibernate). Hibernate, like Spring, can do a lot of dynamic things at runtime. Graal hates that. So we need to get Hibernate to enhance the entities in our application at build time. Add the following Maven plugin to your build.

<plugin>
    <groupId>org.hibernate.orm.tooling</groupId>
    <artifactId>hibernate-enhance-maven-plugin</artifactId>
    <version>${hibernate.version}</version>
    <executions>
        <execution>
            <configuration>
                <failOnError>true</failOnError>
                <enableLazyInitialization>true</enableLazyInitialization>
                <enableDirtyTracking>true</enableDirtyTracking>
                <enableExtendedEnhancement>false</enableExtendedEnhancement>
            </configuration>
            <goals>
                <goal>enhance</goal>
            </goals>
        </execution>
    </executions>
</plugin>

The last thing we'll need to do is to tell Hibernate, at runtime, to not do any enhancement. Create a file, src/main/resources/hibernate.properties:

hibernate.bytecode.provider=none

Now you can compile the application, the same as you did the reactive one, swapping out the main-class. Give it a few minutes. Now you should have two different applications in the target/native-image directory of each application. Run them.

On my machine, the reactive application spins up in 0.106 seconds. The jpa application starts up in 0.181. Fast startup and - the best part - at runtime, these applications will take tens of megabytes, not hundreds (or thousands) as a typical JVM-based application might.

The Next Steps

I can't wait until the Spring Graal 0.8.0 release which will, among other things, baseline on some of the many improvements in Spring Framework 5.3, and possibly include a facility to convert @Configuration-centric Java configurations into Spring's "functional configuration," which requires no proxies or reflection and which is more resource-efficient. I looked at functional configuration in another Spring Tips installment more than three years ago.

Get the Spring newsletter

Stay connected with the Spring newsletter

Subscribe

Get ahead

VMware offers training and certification to turbo-charge your progress.

Learn more

Get support

Tanzu Spring offers support and binaries for OpenJDK™, Spring, and Apache Tomcat® in one simple subscription.

Learn more

Upcoming events

Check out all the upcoming events in the Spring community.

View all