Setup

Requirement

• Ruby 2.2.2 (using 2.5.1 for development)
• Bundler (gem install bundler)
• Docker
• Docker compose
• PostgreSQL (for install gem pg)

Installation

1. $ git clone git@github.com:VLMH/ggpet.git
2. $ cd ggpet
3. $ docker-compose up -d db redis
4. $ bundle
5. $ rails db:create
6. $ rails db:migrate
7. $ rails s

How to test API?

1. Install Postman https://www.getpostman.com/apps
2. Import Postman collection /postman/GGPet.postman_collection

Insert seed data

Run $ rails db:seed This will create 6 pets, 8 customers according to /app-root/db/seeds.rb

Extension 1 - Realtime matching

1. copy .env from email to /app-root/.env
2. $ bundle exec sidekiq
3. $ open /app-root/public/matching.html or open /app-root/public/matching.html in browser
4. Input customer ID and click subscribe
5. Add new pet via Postman
6. You should see new pet appended to the list if the pet matches the customer

There is an issue that the list of customers who adopted pets cannot be updated.

Extension 2 - Geolocation

Since I don't have time for this, I briefly explain how it can be implemented.

Backend

• Add latitude, longitude to Pet
• Get customer location (lat, lon)
• By using gem 'geocoder', query pets near customer (https://github.com/alexreisner/geocoder#for-activerecord-models)
• Make use of /v1/customers/:id/matches and return result

Frontend

• When open matching.html, get permission to retrieve geolocation
• Pass (lat, lon) to API server, when subscribe to a customer
• Display pet locations in map (e.g. Google Map) and result should be sorted by distance

Extension 3 - Scaling

To handle increacing requests, we have to protect the database so that we can scale up server instances.

The first thing to do is to impose repository as a way to access data.

With repository, we can enable the followings with minimum code affected

1. create database read replica
   • handle multiple database connection
2. add caching layer before hit database
   • e.g. get data from Memcache

However, my code extracted services, drivers to separate logic but not repository and is still highly rely on ORM (ActiveRecord). Better do a refactoring before scale up.

Reference: https://blog.lelonek.me/why-is-your-rails-application-still-coupled-to-activerecord-efe34d657c91