AWS, Azure, and GCP each publish a well-architected framework, and beneath the branding they converge on the same idea: every cloud design should be evaluated against a small set of pillars — reliability, security, performance efficiency, cost optimization, and operational excellence — before and after it ships.
Why Pillars Instead of Checklists
Cloud services change weekly, so any product-specific checklist rots fast. Pillars are durable questions: will this design survive component failure, who can access what, does capacity match demand, and what does each request cost? A pillar mindset transfers intact across providers, which is exactly what a pattern-driven architect needs.
Trade-Offs Are the Job
The pillars pull against each other by design. Maximum reliability multiplies cost; maximum security adds latency and friction; maximum performance burns budget. Architecture is the explicit, documented act of choosing where each workload sits on those tensions — a batch reporting job and a payments API deserve very different answers.
Reviews as a Habit
All three vendors structure their frameworks as review questions, and the practice matters more than the paperwork: walk each significant workload through the pillars at design time, then re-review periodically, because both the workload and the platform will have changed underneath it.
Action Step
Pick one workload you know well and score it one-to-five against reliability, security, performance, and cost. Write a single sentence of evidence per score, then note the lowest-scoring pillar — that becomes your running example for the rest of this course.
Educational content only. Architecture decisions depend on your organization’s context and constraints — validate designs against your own requirements and current vendor documentation. This course does not certify anyone.