Common Architectural Patterns for Large-Scale Websites

Each pattern describes a recurring problem in web systems and its core solution, allowing developers to reuse proven approaches without reinventing the wheel.

1. Layered Architecture

Layered architecture divides a system horizontally into distinct parts, each responsible for a simple, single responsibility, and composes them through upper‑to‑lower dependencies. In typical three‑tier web systems the layers are presentation (application), service, and data. Although initially logical, these layers are eventually deployed on separate machines to handle growing traffic.

2. Separation

While layering splits software horizontally, separation splits it vertically. Large sites isolate different functions—shopping, forums, search, advertising—into high‑cohesion, low‑coupling modules, enabling independent development, distributed deployment, and improved concurrency and extensibility.

3. Distributed Deployment

The goal of layering and separation is to distribute modules across multiple servers, increasing CPU, memory, and storage resources to handle higher concurrent access and data volume. Common distributed solutions include distributed applications/services, static asset distribution (static‑resource CDN), distributed storage, and distributed computing frameworks such as Hadoop/MapReduce.

4. Clustering

Modules with concentrated traffic are deployed as server clusters behind load balancers, providing additional concurrency capacity and high availability through failover mechanisms.

5. Caching

Caching reduces server load by returning frequently accessed data directly to users. Implementations include CDN, reverse proxies, local caches, and distributed caches, and are effective when data hotspots exist and cached data remains valid for a short period.

6. Asynchronous Processing

Asynchronous architecture decouples business interactions, breaking operations into stages that communicate via shared data, multithreading, or distributed message queues. A typical pattern is producer‑consumer, where components do not call each other synchronously.

7. Redundancy

To achieve 24/7 availability, redundancy mechanisms such as multiple servers, clustered services, database backups, hot‑cold backups, and geographically distributed disaster‑recovery sites are employed.

8. Automation

Automation spans continuous integration, automated testing, security scanning, deployment pipelines, monitoring, alerting, failover, and recovery, reducing manual effort and improving reliability.

9. Security

Security patterns include authentication via passwords and mobile codes, encrypted communication for login and transactions, CAPTCHAs to deter bots, protection against XSS and SQL injection through encoding, and spam filtering.

Source: http://blog.csdn.net/chaofanwei/article/details/26930627