In this project I will showcase a few service-to-service communications technologies. Most widespread are REST HTTP requests with a client->server direction. More efficient in data intensive services is gRPC. And for Real-time updates we use either websockets (2-way communication) or server-side events (server->client direction).
All these are 1-to-1 communication forms. If we want to notify multiple services (1-to-many communication) we can use events like with RabbitMQ. This also has a retry mechanism so no message gets lost and make the communication asynchronous.
Here we will make a simple REST HTTP request to the server-service. We will use spring boot 3's new HTTP interface which lets us easily declare the client:
public interface RestServerService {
@GetExchange("/test-rest")
ResponseEntity<Post> findTest();
}To spice this example a little bit up, we will 'emulate' a php-service, so sending values in snake-case
and then deserializing properly using @JsonCreator. Furthermore, I like using an UNKNOWN field added as default value
to the enum to avoid serialization issues in the future if a new value was added and not communicated properly.
Google Remote Procedure Call (gRPC) has 3 advantage over classical REST calls:
- serialization of used binary format is faster than JSON
- HTTP/2 support (which has multiplexing, header compression, binary protocol etc.) as compared to HTTP/1 in REST
- strongly typed definitions with Protocol Buffers
So data-intensive services with big payloads are especially benefiting from this. Also, when the services use many different languages the Protocol Buffers are very beneficial.
public interface RestServerService {
@GetExchange("/test-rest")
ResponseEntity<Post> findTest();
}- RabbitMQ instance with dcoker-compsoe (and its new integration!)