The Four Core Thinking Patterns for Software Architecture: Abstraction, Layering, Divide‑and‑Conquer, and Evolution

1. Abstraction

Abstraction is the most important ability in software development and system architecture; it means simplifying a thing by focusing on essential elements while hiding details. The article uses a door and a door handle as visual metaphors and compares software design to building with LEGO blocks, emphasizing top‑down decomposition and bottom‑up composition.

Strong abstract thinkers can see the whole system as a modular castle, breaking it into sub‑modules before implementing each piece.

2. Layered Thinking

Layered architecture divides a complex system into multiple levels, each solving a specific domain problem and providing services to the upper layers. Examples include the classic three‑tier Spring Web application, operating system architecture, and the TCP/IP protocol stack.

Layered thinking also mirrors the evolution of human civilization, where each societal level builds upon the one below.

3. Divide‑and‑Conquer Thinking

Divide‑and‑conquer (or split‑and‑merge) solves complex problems by recursively breaking them into smaller sub‑problems, solving each, and then combining the solutions. The article presents an interview question about sorting a 100 GB file on an 8 GB machine, illustrating how to split the file, sort each chunk, and merge them using external merge sort.

The discussion also highlights recursion as a special case of divide‑and‑conquer and its importance in parser, code‑generator, and serializer implementations.

4. Evolutionary Thinking

Architecture is both designed and evolved; roughly three‑quarters of an internet system’s lifecycle is evolution driven by user feedback. Evolutionary architecture, exemplified by micro‑services, allows continuous adaptation, whereas monolithic systems lack flexibility.

Case studies of companies like eBay, Alibaba, and Netflix show the prevalence of evolutionary paths.

5. How to Cultivate Architectural Thinking

Practical experience, continuous learning, and solid fundamentals from university courses (e.g., data structures and algorithms) lay the groundwork for abstract, layered, and divide‑and‑conquer thinking. The article recommends Berkeley’s CS61B course for strengthening abstraction skills.

Evolutionary design thinking, however, is rarely taught in academia and must be honed through real‑world projects.

Conclusion

1. Architecture’s essence is managing complexity; abstraction, layering, divide‑and‑conquer, and evolution are the four fundamental weapons.

2. Mastering these tools enables design of anything from a single class to a company‑wide platform or organizational architecture.

3. Architecture is dynamic; continuous evolution is required for a system to stay alive.

4. Regularly reflecting on and refining these mental models profoundly changes how architects perceive the world and determines the scale of problems they can solve.

Source: http://www.uml.org.cn/zjjs/201807034.asp