Customers microservice

• Features:

  • Crud API of customers
  • Search endpoint to execute a free text across all customer fields

• Technologies

  • Kotlin
  • Spring
    • Boot
    • MVC
    • Data (Postgres and ElasticSearch)
    • Flyway
    • Cloud Messaging (SNS/SQS)
  • Databases
    • Postgres
    • ElasticSearch
  • LocalStack

• Main Decisions

  1. Monolith first approach
     • We choose to use a monolith first approach with two main packages, because if our project grow we can easily broke in microservices:
       • Package api
         • Have the responsibility to manage all customer operations (Create, Read, Update, Delete, List) in a relational database (Postgres) and produce messages every time that have a change in the customer state.
       • Package searcher
         • Have the responsibility to manage all search operations in a NoSQL database (ElasticSearch), our first search operation is a full text search in all the customers fields
  2. SQS/SNS
     • We choose to use SQS/SNS approach to represent an a async architecture with a component responsible for the search and another with the domain, the best approach right here, is to use a event broker such as Kafka and produce events like CustomerCreated, CustomerUpdated or CustomerDeleted instead of produce a command with the customer data.
  3. LocalStack instead of terraform the infrastructure
     • We choose to use LocalStack instead of terraform because we will need to bootstrap all the infrastructure and it will take a lot of time and a lot of code, because, maybe we will need to build some terraform modules like the ecs because the community module isn't production ready (ECS community module)
  4. Use package by layer
     • Even the package by feature being our favorite approach for microservices we decided to move with a package by layer approach to make more easy to understand the components and the responsibilities.
  5. Functional tests instead of automatized tests :(
     • Due our time to finish the project we didn't have any automatized test for the searcher module, this module had a lot of tests but exclusively functional tests, during this week we will automatized all this tests :)

• How to run locally

  • Prerequisites:
    • git
    • Java v11
    • Docker
    • Docker compose
  • Copy the application.yaml.sample in the src/main/resources Like this:

    cp src/main/resources/application.yaml.sample src/main/resources/application.yaml

  • With the application.yaml created run this make command:

    make local-without-application

  • After the bootstrap of localstack, postgres and elasticsearch you can run the application with the ./gradlew or run the Main class in some IDE

    ./gradlew bootRun