Software architecture is the set of decisions that are expensive to change later: how a system is split into parts, how those parts talk, and where data lives. A good architecture lets a team change one thing without breaking ten others, and it is judged by how the system copes with change, load and failure.
Key concepts
- Modularity: split by responsibility, so each part has one reason to change and a small, stable interface.
- Coupling and cohesion: keep things that change together close, and things that do not, apart.
- Quality attributes: performance, availability, security and changeability pull against each other; name the ones that matter.
- Layers, services, events: a monolith, layered code, services over a network and event-driven flows each trade simplicity for flexibility.
- Decisions on record: write down what was chosen, what was rejected and why.
Vocabulary
- Component: a unit with a defined interface that can be replaced or tested alone.
- Monolith: a system deployed as a single unit.
- Trade-off: a gain in one quality paid for with another.
- ADR: architecture decision record, a short note about one decision.
What to study, in order
- Take a program you wrote and draw its parts and the arrows between them; find the arrow that hurts.
- Read about coupling, cohesion and the dependency rule, then fix one violation in your own code.
- Build the same small app as a monolith and as two services; compare testing, deployment and failure.
- Write an ADR for a real choice, such as a database or a queue.
- Study a long-lived open source system and trace how its structure changed.
Related tags
#design-patterns #api-design #distributed-systems #maintenance