title	README

FitSync - Fitness Tracker App

A Hevy-inspired fitness tracking application built with MongoDB + Redis, developed as a final group project for ENCE614016 — Database System and Laboratory.

---

Group NAMO

Name	Student ID	Role
Nadira Fayyaza Aisy	2406368933	Backend & MongoDB Lead
Naufal Rafif Adigama	2406368965	Redis & Benchmarking Lead
Syifa Sarah Nuraini	2406368883	Frontend & Documentation Lead

Course: ENCE614016 - Database System and Laboratory
Deadline: 15 May 2026

---

Project Overview

FitSync is a fitness tracker app inspired by Hevy. Users can log workouts, track exercise history, view weekly stats, and compete on a streak-based leaderboard. All powered by a dual-database backend.

Why Two Databases?

Database	Role in FitSync	Why
MongoDB	Persistent storage users, workouts, exercises	Document model fits nested workout/exercise data naturally
Redis	Real-time leaderboard, session tokens, API cache	Sub-millisecond reads for live rankings; TTL for sessions

These are not redundant — MongoDB is the source of truth, Redis handles everything time-sensitive and high-frequency.

---

Architecture

Data Flow

---

Project Structure

fitsync/
├── docker-compose.yml          # Spins up API + MongoDB + Redis
├── Dockerfile
├── package.json
├── README.md
├── src/
│   ├── index.js                # Express entry point
│   ├── models/
│   │   ├── User.js             # MongoDB user schema
│   │   └── Workout.js          # MongoDB workout schema (nested exercises)
│   ├── routes/
│   │   ├── auth.js             # Register, Login → Redis SETEX session
│   │   ├── workouts.js         # Log workout → MongoDB + Redis ZADD
│   │   ├── leaderboard.js      # Read top 10 from Redis ZRANGE
│   │   └── stats.js            # Weekly stats (cache-aside pattern)
│   └── middleware/
│       └── auth.js             # JWT + Redis session validation
├── scripts/
│   └── seed.js                 # Populates MongoDB + Redis with sample data
├── benchmarks/
│   ├── benchmark.js            # Latency comparison: Redis cache vs MongoDB
│   ├── results.csv             # Raw benchmark output
│   └── benchmark_plot.png      # Bar chart of results
└── docs/
    ├── data_models.md          # MongoDB schemas + Redis key design
    ├── architecture.png        # Architecture diagram
    └── design_decisions.md     # Why we chose each DB for each role

---

Quick Start

Prerequisites

• Docker Desktop installed and running

1. Clone the repository

git clone https://github.com/<your-username>/fitsync-namo.git
cd fitsync-namo

2. Start all services

docker compose up --build

This starts three containers: fitsync_api (port 3000), fitsync_mongo (port 27017), fitsync_redis (port 6379).

3. Seed sample data

docker exec fitsync_api node scripts/seed.js

This creates 3 users (nadira, naufal, syifa), 15 workouts, and seeds the Redis leaderboard.

4. Test the API

GET  http://localhost:3000/
GET  http://localhost:3000/api/leaderboard
POST http://localhost:3000/api/auth/login
POST http://localhost:3000/api/workouts
GET  http://localhost:3000/api/stats/weekly/:userId

---

API Reference

Auth

Method	Endpoint	Body	Description
POST	/api/auth/register	{ username, email, password, profile }	Register new user → saved to MongoDB
POST	/api/auth/login	{ email, password }	Login → session token stored in Redis SETEX

Workouts

Method	Endpoint	Auth	Description
POST	/api/workouts	✅	Log workout → MongoDB + updates Redis leaderboard
GET	/api/workouts	✅	Get all workouts for current user from MongoDB

Leaderboard & Stats

Method	Endpoint	Auth	Description
GET	/api/leaderboard	❌	Top 10 streak rankings → read directly from Redis
GET	/api/stats/weekly/:userId	✅	Weekly stats → Redis cache first, MongoDB on miss

---

Data Models

MongoDB — users collection

{
  "_id": "ObjectId",
  "username": "nadira",
  "email": "nadira@fitsync.com",
  "password_hash": "...",
  "created_at": "2026-04-26T00:00:00Z",
  "profile": {
    "age": 21,
    "weight_kg": 55,
    "height_cm": 162
  }
}

MongoDB — workouts collection

{
  "_id": "ObjectId",
  "user_id": "ObjectId",
  "date": "2026-05-05T10:00:00Z",
  "duration_min": 45,
  "calories_burned": 320,
  "exercises": [
    { "name": "Squats", "sets": 4, "reps": 12, "weight_kg": 40 },
    { "name": "Deadlift", "sets": 3, "reps": 8, "weight_kg": 80 }
  ]
}

Redis Key Design

Key Pattern	Type	Command	TTL
leaderboard	Sorted Set	ZADD leaderboard <streak> <userId>	No expiry
session:<token>	String	SETEX	3600s (1 hr)
cache:weekly:<userId>	String/JSON	SETEX	300s (5 min)

---

Benchmarks

Benchmarks compare Redis cache read vs MongoDB aggregation query for the weekly stats endpoint.

Results saved in /benchmarks/results.csv. See /benchmarks/benchmark_plot.png for the chart.

To run benchmarks yourself:

docker exec fitsync_api node benchmarks/benchmark.js