Why Data Movement Is the Core of Internet Layered Architecture

21CTO Community Introduction: This article is authored by architect Shen Jian of 58.com. It is concise and aims to provide value.

The diagram above shows a typical Internet layered architecture:

Client layer : typically a browser or an app.

Site application layer : implements core business logic, fetches data from downstream services, and returns HTML or JSON upstream.

Data‑cache layer : accelerates data access.

Data‑database layer : persists data.

If the system is service‑oriented, an additional service layer appears between the application and data layers:

Within the same tier, such as an app or a web server, the classic MVC separation also applies:

View layer : presentation.

Control layer : business logic.

Model layer : data representation.

Every engineer implicitly implements this layered architecture. The essence of the Internet layered architecture is a three‑step process: data movement , then processing , and finally presentation . Data movement is the core of the entire flow.

Data processing and presentation require CPU computation, and the CPU is fixed:

Databases, services, and web servers run on fixed clusters.

Clients (browsers or apps) also have fixed CPUs.

Thus, data is the moving part:

Cross‑process movement : data travels from database/cache → service layer → web‑server layer → client layer.

In‑process movement : data moves from model → control → view.

Because data moves, two aspects are crucial:

Data transmission format (protocol).

Data shape at each layer.

Transmission formats include binary or text protocols between service and DB/cache, RPC binary protocols between web‑server and service, and HTTP between client and web‑server.

Considering user data as an example, its shape evolves through the layers:

In the DB layer, data exists as rows (uid, name, age).

In the cache layer, data is stored as key‑value pairs (uid → User).

The service layer converts rows or KV pairs into a program‑friendly User object.

The web‑server layer transforms the User object into a JSON object suitable for HTTP.

The client finally receives the JSON object.

Conclusion: The essence of Internet layered architecture is data movement.

Understanding this principle and the associated encapsulation and reuse methods helps answer common questions such as when to introduce a DAO layer, when to adopt service‑orientation, when to extract a common middle‑platform, and when to separate front‑end and back‑end.

Further details on architecture evolution will be explored in the next article.

Summary

The core of Internet layered architecture is data movement.

Transmission protocols and data shapes at each layer are critical.

Evolution principles focus on encapsulation and reuse.

Think: Which system’s architecture does not follow “fixed CPU, moving data” but instead “fixed data, moving CPU”?