@ -10,25 +10,25 @@ All those services will be in turn encapsulated in different virtual machines.
#### General overview
We will setup a movie streaming platform, where one API (`inventory`) will have information on the movies available and another one (`orders`) will process the payments.
We will setup a movie streaming platform, where one API (`inventory`) will have information on the movies available and another one (`billing`) will process the payments.
The API gateway will communicate in HTTP with `inventory` and using RabbitMQ for `orders`.
The API gateway will communicate in HTTP with `inventory` and using RabbitMQ for `billing`.
In this exercise you will need to install Node.js (with Express, Sequelize and other packages), PostgreSQL and RabbitMQ.
In this exercise you will need to install Node.js (with Express, Sequelize and other packages), PostgreSQL, RabbitMQ, Postman, VirtualBox and Vagrant.
While it may seems overwhelming at first there is a lot of resources available both on official website and on community blogs about setting up those tools.
Also the specific configuration details may change from platform to platform so don't hesitate to play around with it and be sure to check everything is installed correctly before to move on.
Also the specific configuration details may change from platform to platform so don't hesitate to play around with it and be sure everything is installed correctly before to move on.
#### API 1: Inventory
##### Definition of the API
##### Definition of the Inventory API
This API will be a CRUD (Create, Read, Update, Delete) RESTful API.
It will use a PostgreSQL database.
It will provide information about the movies present in the inventory and allow users to do basic operations on it.
A common way to do so is to use Express (TODO add a link?) which is a popular Node.js web framework.
We will couple it with Sequelize (TODO add a link?), an ORM (TODO add a link?) which will abstract and simplify the interactions between our API and the database.
A common way to do so is to use Express which is a popular Node.js web framework.
We will couple it with Sequelize, an ORM which will abstract and simplify the interactions between our API and the database.
Here are the endpoints with the possible HTTP requests:
@ -54,7 +54,9 @@ The API should work on `http://localhost:8080/`.
##### Defining the Database
The database will done with PostgreSQL and it will be called `movies`.
For the database we will use PostgreSQL.
The database will be called `movies`.
Each movie will have the following columns:
- `id`: autogenerated unique identifier.
@ -63,11 +65,11 @@ Each movie will have the following columns:
(TODO add link on how to setup it?)
##### Testing the API
##### Testing the Inventory API
In order to test the correctness of your API you should use Postman (TODO add a link?). You could create one or more tests for every endpoint and then export the configuration so you will be able to reproduce the tests on different machines easily.
#### API 2: Orders
#### API 2: Billing
#### The API Gateway
@ -75,7 +77,7 @@ The Gateway will be the only service accessible by the user, it will take care o
The API Gateway should work on `http://localhost:3000/`.
##### Interfacing with API1
##### Interfacing with Inventory API
The gateway will route all requests to `/api/movies` at the API1, without any need to check the information passed through it.
It will return the exact response received by the API1.
@ -83,16 +85,19 @@ It will return the exact response received by the API1.
In order to achieve this goal it will be necessary to setup a proxy system.
You could check `http-proxy-middleware` npm package in order to achieve such goal.
##### Interfacing with API2
##### Interfacing with Billing API
The gateway will receive POST requests from `api/orders` and send a message using RabbitMQ in a queue called `task_queue`.
The gateway will receive POST requests from `api/billing` and send a message using RabbitMQ in a queue called `billing_queue`.
The content of the message will be the POST request body stringified with `JSON.stringify`.
The Gateway should be able to send messages to the API2 even if that API is not running. When the API2 will be started it should be able to process that message and send an acknowledgement back.
##### Documenting the API
Good documentation is a very critical feature of every API. By design the APIs are meant for others to use, so there have been very good efforts to create standard and easy to implement ways to document it.
As an introduction to the art of great documentation you must create an OpenAPI documentation file for the API Gateway. There is many different ways to do so, a good start could be using SwaggerHub (TODO add a link?) with at least a meaningful description for each endpoint. Feel free to implement any further and more complex feature.
As an introduction to the art of great documentation you must create an OpenAPI documentation file for the API Gateway. There is many different ways to do so, a good start could be using SwaggerHub with at least a meaningful description for each endpoint. Feel free to implement any extra feature as you see fit.
#### Virtual Machines
#### Overall file structure
@ -100,34 +105,38 @@ You can organize your internal file structure as you prefer. That said here is a
```console
.
├── inventory
│ ├── app
│ │ ├── config
│ │ ├── controllers
│ │ ├── models
│ │ └── routes
│ ├── node_modules
│ ├── package.json
│ ├── package-lock.json
│ └── server.js
├── orders
│ ├── app
│ │ ├── config
│ │ ├── controllers
│ │ └── models
│ ├── node_modules
│ ├── package.json
│ ├── package-lock.json
│ └── server.js
└── api-gateway
├── node_modules
├── package.json
├── package-lock.json
├── proxy.js
├── routes.js
└── server.js
├── config.yaml
├── .env
├── scripts
│ └── [...]
├── srcs
│ ├── api-gateway
│ │ ├── package.json
│ │ ├── proxy.js
│ │ ├── routes.js
│ │ └── server.js
│ ├── billing-app
│ │ ├── app
│ │ │ ├── config // This is a directory with some .js files
│ │ │ ├── controllers // This is a directory with some .js files
│ │ │ └── models // This is a directory with some .js files
│ │ ├── package.json
│ │ └── server.js
│ └── inventory-app
│ ├── app
│ │ ├── config // This is a directory with some .js files
│ │ ├── controllers // This is a directory with some .js files
│ │ ├── models // This is a directory with some .js files
│ │ └── routes // This is a directory with some .js files
│ ├── package.json
│ └── server.js
└── Vagrantfile
```
You should be able to start the API Gateway and the two APIs by using the command `node server.js` inside their respective directories.
The Gateway should be able to send messages to the API2 even if that API is not running. When the API2 will be started it should be able to process that message and send an acknowledgement back.
If you decide to use a different structure for your project remember you should be able to explain and justify your decision during the audit.
For clarity autogenerated files and directories like `node_modules` and `package-lock.json` have not been included in that tree.
> As a best practice it is strongly advised to add `node_modules` to your `.gitignore` in order not to upload useless files into your git repository (they will be autogenerated during the build process).
Michele Sessa
2 years ago
committed by