How a Five‑Tier Cache Boosts E‑Commerce Product Retrieval Speed

0 Introduction

In e‑commerce systems, fast retrieval of product information is crucial for user experience. This article explains a multi‑level cache architecture designed to improve performance and reliability.

1 Overall Architecture

The cache system uses a five‑tier strategy from the front‑end to the back‑end:

CDN cache

Nginx cache cluster

Redis cache

JVM local cache

MySQL persistent storage

2 Detailed Walk‑through

1. User Request Entry

Requests first hit the CDN, which serves static assets and reduces load on origin servers.

User -> CDN -> Front‑end Static Resources

2. Load Balancing

After CDN, traffic passes through LVS and HAProxy for high‑performance, high‑availability load balancing. CDN -> LVS (Linux Virtual Server) + HA Proxy LVS provides high‑performance load distribution.

HA Proxy adds fine‑grained traffic control and health checks.

3. Nginx Edge (JavaEdge)

JavaEdge is an Nginx cluster that performs initial request handling and forwarding, including IP rate‑limiting and URL parsing.

LVS + HA Proxy -> JavaEdge (Nginx forwarding layer)

IP throttling and forwarding

Business routing based on URL

4. Nginx Business Layer

Requests are routed to specific business Nginx instances such as product‑page or checkout Nginx.

JavaEdge -> Single‑product Nginx / Checkout Nginx

5. Lua Scripts & Redis Cache

Lua scripts inside Nginx interact with Redis to fetch cached data directly.

Nginx Business Layer -> Lua -> Redis

6. JVM Cache

If Redis misses, the request falls back to a JVM‑level local cache for faster access.

Redis (miss) -> JVM Cache

7. MySQL Persistence

When all caches miss, the system queries MySQL and then updates the upper caches.

JVM Cache (miss) -> MySQL

3 Cache Levels

The five layers are:

Level 1: CDN or browser cache

Level 2: Nginx cache

Level 3: Redis cache

Level 4: JVM local cache

Level 5: MySQL as the ultimate data source

4 Special Notes

Nginx local cache stores hot data to speed up access.

Redis master‑slave sync ensures high availability.

JVM‑to‑Redis updates keep data consistent across layers.

5 Conclusion

The five‑tier cache architecture leverages CDN, Nginx, Redis, JVM cache and MySQL to deliver rapid product information retrieval, improve system performance, reliability and scalability. In practice, one must also handle cache consistency, expiration policies and hot‑data management.