Design System of High-Scale Product Application(POC)

Given a real-world problem to scale the applications to read and save products inside a controlled environment I decided to create a POC (Prove of Concepts) about the approach to delivering a good performance and high availability design.

This application backend is responsible by save and read products over the HTTP protocol using a RESTFull approach, the user makes a read or write request, the service does processing, stores data, then returns the results.

The service needs to envolve from serving a small amount of requests to millions of requests.

How use

Its a POC that you can test it in your development enviroment. To init the application you need to have a docker and docker-compose installed in your computer.

docker-compose up -d

With the all containers running you can call the request to use the application:

Create a Product

To create a new product execute a POST in our API:

Schema

{
  "description": "string",
  "value": "float"
}

Execute the below command to create with 201 status:

curl -X POST -H "Content-Type: application/json" -d '{ "description": "Awesome Product", "value": 10.0}' http://localhost/products

Get all Products

Route only available to get all:

curl -X GET http://localhost/products

Returns 200 status:

{
  "products": [
    {
      "description": "Awesome Product",
      "id": 6,
      "tax": 12.5,
      "value": 12.5
    }
  ]
}

Create a Comment

To create a new comment to product execute a POST in our API:

Schema

{
  "comment": "string",
  "author": "string",
  "id_product": "int"
}

Execute the below command to create with 201 status:

curl -X POST -H "Content-Type: application/json" -d '{ "id_product": 1, "comment": "Hello!!!", "author": "Hattori" } ' http://localhost/comments

High Level Design

• Client using a brower, mobile, or any device to access to application
• DNS translate a domain name to an IP address
• Web Server is a proxy that centralizes internal services and provides unifield inrterfaces
• The Application Servers Product_Write and Product_Read handling the requets, with this separation we can scale both layers independently, the Product_Write is reponsable for insert new data while Product_Read read all data from storage.
• Tax is a small microservice responsable by calculate all tax to some product, as a internal service that use over gRPC protocol to communicate with others.
• The database choised was the Postgres and Redis as cache the Comments_Worker uses as queue broker too, in production enviroment it's not a good approach.
• Comments_Write_Async is a bach service that is responsable by create asynchronously Comments, the client send request to async API than pubisher messase to Redis broker, the Comments_Worker receive message and process to save to No-SQL database(MongoDB)