Design Scalable Website Architecture: Clusters, Load Balancing & Consistent Hashing

1. Scalability Design of Website Architecture

Physical separation of different functions—vertical separation of business processing stages and horizontal separation of independent business modules—enables system scalability.

Single‑function scaling is achieved by deploying server clusters. Application‑server clusters and data‑server clusters have distinct scaling characteristics.

2. Application Server Cluster Scalability Design

Application servers should be stateless so any request can be processed by any server in the cluster; HTTP is inherently stateless, with state managed via cookies or sessions.

Load balancing distributes HTTP requests across the cluster. Common methods include:

HTTP redirect (simple but adds latency and potential SEO issues)

DNS resolution (offloads balancing to DNS but suffers from caching delays)

Reverse‑proxy balancing (integrated with proxy servers, easy to deploy but may become a bottleneck)

IP‑level balancing (high performance in kernel space, limited by network‑card bandwidth)

Data‑link‑layer balancing (direct routing, minimal latency, widely used via Linux LVS)

3. Distributed Cache Cluster Scalability Design

Cache clusters cannot rely on simple load balancing because cached data is partitioned; the goal is to add new cache nodes with minimal impact on existing data.

Using Memcached as an example, a consistent‑hash ring maps keys to cache nodes. When a new node joins, only keys that fall into its segment are remapped, dramatically reducing cache‑miss rates compared with ordinary modulo hashing.

4. Data Storage Server Cluster Scalability Design

Relational databases achieve scaling through replication (master‑slave) for read‑heavy workloads and sharding (data partitioning) for horizontal scaling, though cross‑shard joins are not possible.

NoSQL databases prioritize high availability and scalability; HBase, for instance, uses region splitting and the HDFS distributed file system to grow horizontally.

5. Summary

Achieving website scalability hinges on well‑designed server clusters and appropriate scaling techniques; while no universal perfect solution exists, thoughtful architecture allows new servers to be added with minimal or no impact on existing services.

References

Li Zhihui, "Large‑Scale Website Architecture – Core Principles and Case Studies"

Laoxu, "HTTP Stateless Protocol and Session Principles"

Baidu Baike, "Consistent Hash Algorithm"

Charlee, "Complete Analysis of Memcached"

Bluishglc, "Database Sharding Basics and Strategies"