Här finns tekniska artiklar, presentationer och nyheter om arkitektur och systemutveckling. Håll dig uppdaterad, följ oss på Twitter

Callista medarbetare Anna Eriksson

R2DBC - Reactive Programming with Spring, Part 4.

// Anna Eriksson

This is part four of my blog series on reactive programming, which will give an introduction to R2DBC and describe how we can use Spring Data R2DBC to create a fully reactive application.

1. What is R2DBC?

If you are not already familiar with reactive programming and Reactive Streams, I recommend you to first read my introduction on reactive programming which describes the motivation behind this programming paradigm.

When developing a reactive application that should include access to a relational database, JDBC is not a good fit, since it is a blocking API.

R2DBC stands for Reactive Relational Database Connectivity and is intended to provide a way to work with SQL databases using a fully reactive, non-blocking API. It is based on the Reactive Streams specification and is primarily an SPI (Service Provider Interface) for database driver implementors and client library authors - meaning it is not intended to be used directly in application code.

At this moment, driver implementations exist for Oracle, Microsoft SQL Server, MySQL, PostgreSQL, H2, MariaDB and Google Cloud Spanner.

2. Spring Data R2DBC

Spring Data offers an R2DBC client - Spring Data R2DBC.

This is not a full ORM like JPA - it does not offer features such as caching or lazy loading. But it does provide object mapping functionality and a Repository abstraction.

To demonstrate how it can be used, let’s revisit the StudentController example from my previous blog on WebFlux:

public class StudentController {

    private StudentService studentService;

    public StudentController() {

    public Mono<ResponseEntity<Student>> getStudent(@PathVariable long id) {
        return studentService.findStudentById(id)

    public Flux<Student> listStudents(@RequestParam(name = "name", required = false) String name) {
        return studentService.findStudentsByName(name);

    public Mono<Student> addNewStudent(@RequestBody Student student) {
        return studentService.addNewStudent(student);

    public Mono<ResponseEntity<Student>> updateStudent(@PathVariable long id, @RequestBody Student student) {
        return studentService.updateStudent(id, student)

    public Mono<ResponseEntity<Void>> deleteStudent(@PathVariable long id) {
        return studentService.findStudentById(id)
                .flatMap(s ->
                                .then(Mono.just(new ResponseEntity<Void>(HttpStatus.OK)))
                .defaultIfEmpty(new ResponseEntity<>(HttpStatus.NOT_FOUND));

This controller holds some different methods for performing actions on students. We can see that it is using a StudentService to perform these actions. Now we will look into this functionality behind the REST controller and how we can implement database access using R2DBC.

2.1 Implementation example

2.1.1 Dependencies

First, we need to add a couple of new dependencies to our project:



We need to include the spring-boot-starter-data-r2dbc to enable spring-data-r2dbc. For this example we will use a postgresql database, and so we need to add the r2dbc-postgresql to get the r2dbc driver implementation needed.

2.1.2 Database config

We can either add our database connection details in


or use a Java-based configuration:

import io.r2dbc.spi.ConnectionFactories;
import io.r2dbc.spi.ConnectionFactory;
import io.r2dbc.spi.ConnectionFactoryOptions;
import org.springframework.context.annotation.Bean;
import org.springframework.context.annotation.Configuration;

import static io.r2dbc.spi.ConnectionFactoryOptions.*;

public class R2DBCConfig {

    public ConnectionFactory connectionFactory() {
        return ConnectionFactories.get(
                        .option(DRIVER, "postgresql")
                        .option(HOST, "localhost")
                        .option(USER, "user")
                        .option(PASSWORD, "secret")
                        .option(DATABASE, "studentdb")


2.1.3 StudentService

Now let’s take a look at the StudentService that the StudentController is using:

public class StudentService {

    private StudentRepository studentRepository;

    public StudentService() {

    public Flux<Student> findStudentsByName(String name) {
        return (name != null) ? studentRepository.findByName(name) : studentRepository.findAll();

    public Mono<Student> findStudentById(long id) {
        return studentRepository.findById(id);

    public Mono<Student> addNewStudent(Student student) {

    public Mono<Student> updateStudent(long id, Student student) {
        return studentRepository.findById(id)
                .flatMap(s -> {


    public Mono<Void> deleteStudent(Student student) {
        return studentRepository.delete(student);


As you can see, it uses a StudentRepository to perform the different database operations on students. So now let’s take a look at this repository.

2.1.4 StudentRepository

The StudentRepository is an implementation of ReactiveCrudRepository. This is an interface from Spring Data R2DBC for generic CRUD operations using Project Reactor types. Since ReactiveCrudRepository already holds definitions for most of the repository methods we use in the StudentService (findAll, findById, save and delete) all we need to declare is the following:

public interface StudentRepository extends ReactiveCrudRepository<Student, Long> {

    public Flux<Student> findByName(String name);


More complex queries could be defined as well by adding a @Query annotation to a method and specifying the actual sql.

Besides the ReactiveCrudRepository, there is also an extension called ReactiveSortingRepository which provides additional methods to retrieve entities sorted.

2.1.5 Student

Finally, let’s look at the implementation of Student:

import lombok.AllArgsConstructor;
import lombok.Data;
import lombok.NoArgsConstructor;

public class Student {

    private Long id;
    private String name;
    private String address;


A few things to note:

  • The id of an entity should be annotated with Spring Data’s @Id annotation.

  • The @Table annotation is not necessary but adding it lets the classpath scanner find and pre-process the entities to extract the related metadata. If you don’t add it this will instead happen the first time you store an entity which could have a slightly negative impact on performance.

  • Lombok is recommended to be used to avoid boilerplate code.

  • There are also some other recommendations to ensure you get optimal performance, you can find the details in the reference documentation.

2.1.6 Other options for queries

Instead of using a repository, you could execute an SQL statement directly using a DatabaseClient.

For example, to retrieve all students:

 public Flux<Student> findAll() {
        DatabaseClient client = DatabaseClient.create(connectionFactory);
        return client.sql("select * from student")
                .map(row -> new Student(row.get("id", Long.class),
                        row.get("name", String.class),
                        row.get("address", String.class))).all();

It is also possible to use the R2dbcEntityTemplate to perform operations on entities. For example:

private R2dbcEntityTemplate template;

public Flux<Student> findAll() {

public Mono<Void> delete(Student student) {
    return template.delete(student).then();

2.2 Other features

2.2.1 Optimistic locking

Quite similar to JPA, it is possible to apply a @Version annotation at field level, to ensure that updates are only applied to rows with a matching version - if the version is not matching an OptimisticLockingFailureException is thrown.

2.2.2 Transactions

Spring supports reactive transaction management through the ReactiveTransactionManager SPI. The @Transactional annotation can be applied on reactive methods returning Publisher types and programmatic transaction management can be applied using the TransactionalOperator.

2.2.3 Reactive libraries

Just like WebFlux, Spring Data R2DBC requires Project Reactor as a core dependency but is interoperable with other reactive libraries that implement the Reactive Streams specification. Repositories exist for RxJava2 and RxJava3 as well (view package summary).

2.2.4 Connection pooling

For connection pooling, there is a library called r2dbc-pool available. For details on how to use it, take a look here.

3. Production readiness

R2DBC is still a fairly new technology. The latest release versions as of now:

  • R2DBC specification: 0.8.5
  • Spring Data R2DBC: 1.3.1
  • r2dbc-postgresql: 0.8.8
  • r2dbc-pool: 0.8.7

Before deciding to go to production with this for your application, it is of course recommended to take a closer look at the current state of the database driver and pooling implementations compared to your requirements. There are some open issues that might prevent you from taking this step as of now, but improvements are ongoing.

4. To summarize…

This blog post demonstrated how Spring Data R2DBC can be used in a WebFlux application. And by that, we have created a fully reactive application and also come to an end of this series on reactive programming.

Another very interesting initiative worth mentioning is Project Loom. This is an OpenJDK project that started already in 2017 aiming to deliver lightweight concurrency, including a new type of Java threads that do not directly correspond to dedicated OS threads. This type of virtual threads would be much cheaper to create and block.

As you might recall from my first blog post the key drivers behind the reactive programming model is that we:

  • move away from the thread per request model and can handle more requests with a low number of threads
  • prevent threads from blocking while waiting for I/O operations to complete
  • make it easy to do parallel calls
  • support “back pressure”, giving the client a possibility to inform the server on how much load it can handle

Project Loom seems very promising when it comes to helping out with the first two items in this list - this would then be taken care of by the JVM itself without any additional framework needed.

It is not yet decided when the changes will be introduced in an official Java release, but early access binaries are available for download.



Spring Data R2DBC Reference Documentation



Project Loom

Going inside Java’s Project Loom and virtual threads

Tack för att du läser Callistas blogg.
Hjälp oss att nå ut med information genom att dela nyheter och artiklar i ditt nätverk.