This project is a prototype, coded with no much time available, capable of handling a large number of trade messages per second by receiving them through a HTTP endpoint, which takes JSON as payload. By prototype I mean something functional in processing lots of trade, but generally some things act naively.
This project was built using microservices approach on top of Java, which leverages some well known open-source technologies like Spring, Hibernate, Eureka, Zuul, Feign, OAuth2, Maven, JUnit, Mockito, Gatling, Lombok, Mysql, Docker compose.
Some of the main characteristics of a microservices-based application are functional decomposition, well-defined interfaces and single resposibility. In other words, microservice allows us to get loose coupling, high cohesion, agility, flexibility and scalability.
Here're some patterns related to microservice:
With this in mind the trade processor was split into the following Maven modules:
Inside each module you can find a brief doc around them.
- Java 8
- Maven 3
- Docker compose 1.8
- Fork the project
- Build:
mvn install - Before running docker make sure all exposed ports in the docker compose file are really available
- Run docker-compose to start up all services together:
docker-compose up - Allow some seconds until all services are running and properly registered within Service Discovery
Now you've got everything up and running you must authenticate in order to be able to call the main entrypoint. With the following info you must be able to get the OAuth2 token:
- Username: demo
- Password: demo
- Endpoint: http://docker-host:8080/uaa/oauth/token
Using Postman to get access token:
Optionally you can get the load-testing Maven project to inject random trade data using the following command:
mvn gatling:test -Doauth2.token=YOUR-GENERATED-OAUTH-TOKEN
Remember to do the security step to get a valid oauth token.
http://docker-host:8080/consumption/trade (main entry point)
{
"userId" : 134256,
"currencyFrom" : "EUR",
"currencyTo" : "GBP",
"amountSell" : 1000.0,
"amountBuy" : 747.1,
"rate" : 0.7471,
"timePlaced" : "24-JAN-15 10:27:44",
"originatingCountry" : "FR"
}http://docker-host:8080/consumption/trades (batch consume)
http://docker-host:8080/processing/trade (simple http entry point that exposes stored trade messages)
ws://docker-host:8080/processing/real-time (websocket entry point that exposes the latest metrics)
ws://docker-host:8080/processing/real-time/history (websocket entry point that exposes metric history)
Note that all entry points are taking into account the API Gateway port.