Architectural Design Basics: High‑Performance, High‑Availability, and Scalability Patterns

This article extracts the core ideas from the book "From Zero to Architecture" and combines personal experience to present practical methods for designing high‑performance, high‑availability, and scalable systems.

Basic Concepts and Design Methods – Clarifies terminology such as system, subsystem, module, component, framework, and architecture, emphasizing that architecture is the concrete implementation of a set of development guidelines.

Complexity Sources – Identifies performance, availability, and scalability as the main dimensions of system complexity and explains how to recognize the specific challenges a business faces.

High‑Performance Architecture – Discusses storage performance (read/write separation, master‑slave replication, handling replication lag) and compute performance (single‑machine optimizations, multi‑process/threading, asynchronous I/O). It also covers load‑balancing strategies (DNS, hardware, software) and data sharding techniques (range, hash, configuration routing).

NoSQL and Caching – Introduces NoSQL databases (key‑value, document, column‑family, search) as solutions for specific performance or schema problems, and explains cache benefits and pitfalls (cache penetration, snowball, hot‑key issues) along with mitigation techniques.

High‑Availability Architecture – Explains CAP and BASE theories, failure‑mode analysis (FMEA), and business continuity planning (BCP). It details storage HA patterns (master‑backup, master‑slave, active‑active, double‑machine failover modes) and compute HA patterns (active‑passive, active‑active, symmetric/asymmetric clusters). It also describes multi‑region active‑active deployments and data‑center partitioning strategies.

Interface‑Level Fault Tolerance – Covers pre‑emptive measures (rate limiting, queuing) and reactive measures (circuit breaking, degradation) with algorithms such as fixed/sliding windows, token bucket, and leaky bucket.

Scalable Architecture – Emphasizes the need for modularity and separation of concerns. Presents layered architecture, SOA, micro‑services, and micro‑kernel (plugin) approaches, including guidelines for service decomposition, stability‑based splitting, and performance‑based splitting.

Evolution and Refactoring – Advises a gradual, principle‑driven evolution: start simple, apply the "appropriate", "simple", and "evolutionary" principles, and refactor only after encountering recurring patterns.

The article concludes with a roadmap for architects: identify the most critical complexities, prioritize low‑cost high‑impact improvements, iterate in 1‑3 month cycles, and collaborate with upstream/downstream teams to achieve sustainable, high‑quality system design.