FlashGuard

A high performance distributed inventory locking system designed to prevent race conditions in high-traffic e-commerce environments.

Tech stack

Spring Boot 3, AWS(S3), Redis, Next js 15, PostgresSQL(supabase), hosting - render, vercel

Technical Problem

• The Problem: Traditional database locking (PostgreSQL) causes deadlocks and high latency when 10,000+ users try to buy a limited-stock item simultaneously.
  

• The Solution: Implemented Redis Lua scripts to handle inventory validation atomically in-memory before persisting data to the database.

Architecture Diagram

Key features & Performance Metrics

• High concurrency - optimized to handle 10k+ requests per second. (RPS)
  
• Deadlock Reduction - Decoupled validation from persistence, slashing DB contention by 80%.
  
• Atomic Locking - Used Redis Lua scripts to ensure "Exactly-once" processing for inventory updates.
  
• CI/CD & Security - Automated testing with GitHub Actions and security scanning with CodeQL.
  

Live Demo

• Frontend (Next.js): flash-guard.vercel.app
• Backend (Spring Boot): flashguard.onrender.com

RoadMap

[x] core redis locking logic.
[x] spring boot API integration.
[x] AWS S3 image hosting.
[x] Advanced analytics dashboard.

API Testing & Documentation

The API is fully documented and tested using Posstman.

• Postman Collection: You can find the raw JSON collection here
• How to test:
• 1. Import the collection into postman.
  2. Set the base_url .
  3. Run the "Stress Test" folder ot simulate inventory lockinng.

Sample API Response (Locking Success)

{
    "oversold": 0,
    "successfulLocks": 8,
    "avgLatency": 54.0939,
    "productId": 6,
    "totalRequests": 10000
}

CI/CD & Security

• 
• 
• 

Testing

• Unit tests: JUnit & Mockito for service logic.
  
• Stress testing: simulated 10k concurrent users to validate redis performance.