Event Ticket Reservation System

Backend hosted at https://www.ticketeer.ml/api/v1
Frontend hosted at www.ticketeer.ml

Objectives:

• [Main] Learn and familiarize with Go programming language
• [Main] Concurrency in practice (with and w/o database features help)
• Pinpoint possible limitations in each implementation phase’s design
• Design for optimal performance (where to store data and in what form it should be stored)

Functional Requirements:

1. User System Functionalities
   • Register / Sign-up
   • Login
   • User Roles
   • Admin
   • Event Organizer
   • Customer
2. Event Management Functionalities
   • Create/View/Edit/Delete Event
   • Required Properties
   • Event Name
   • Ticket Quota/Limit

• Admins can view and delete all events in the system
• Event Organizers can create/view/edit/delete their own events
• Customers can view all events and are able to make reservation(s) on any event. (Customers can reserve more than 1 ticket per request.)
• Tickets can only be sold if there’s quota left or have not reached their limits
• Customers are able to cancel their reservations
• Event Organizers can see total ticket reserved / remaining ticket quota

Design and implement a (backend) service with HTTP APIs serving the above mentioned functionalities using Go programming language

Phase 1 - Without using persistence (database). Store all data in the program's memory.

Phase 2 - Use persistence (relational database) (e.g. PostgreSQL) to store quota (naive implementation)

Phase 3 - Optimize and/or redesign for performance

Prerequisites

• Go 1.13 or later

1. Start PostgreSQL

   docker run --rm -p 5432:5432 -e POSTGRES_PASSWORD=postgres --name ticket_reservation_postgres postgres:12-alpine

2. Create database

   docker exec -i ticket_reservation_postgres psql -U postgres -c "drop database if exists ticket_reservation" &&
   docker exec -i ticket_reservation_postgres psql -U postgres -c "create database ticket_reservation"

3. Create tables and run server

   go run main.go migrate-db
   go run -tags debug main.go serve-api

4. Connect to psql

   docker exec -ti ticket_reservation_postgres psql -U postgres

Performance Test

Uses autocannon with config:

1. connections: 80
2. duration: 10
3. 4 POST requests. Each request reserves 1 ticket from 1 event
4. excludeErrorStats: true

Start by sending a GET request to populate DB (ONCE) with

• EventID 1 Quota 10000
• EventID 2 Quota 10000
• EventID 3 Quota 10000
• EventID 4 Quota 10000

curl localhost:9092/api/v1/pop

Then run load_test/cannon.js with updated customer tokens
P.S. Need tokens because my implementation uses jwtclaims to get user data 🤣

Results

Phase	Requests Sent	Avg Latency	Avg Throughput
1 Memory only	9307	85.97	300800
2 Postgres only (with retry)	713	513.71	10302.21
3 Postgres with batch jobs in memory	4992	161.47	119737.6
4 Postgres + Redis	2958	265.96	62724

Docker useful commands

1. docker build -t ticket_reservation_server .
2. docker run --rm -ti -p 9092:9092 ticket_reservation_server serve-api
3. docker run --rm -ti --entrypoint="/bin/bash" ticket_reservation_server