Two boundaries between pillars do most of the discriminating work. Reliability and Performance Efficiency both involve scaling, and they separate on why. Scaling to survive a failure or a surge is Reliability; picking the right instance family for the work is Performance Efficiency. Cost Optimization and Sustainability point the same way most of the time, since an idle instance wastes both money and energy. They separate on what is being minimised: spend or environmental impact.
Operational Excellence is the pillar people under-recognise, because its vocabulary sounds like process rather than architecture. Observability, safe automation, frequent small reversible changes and learning from operational events all sit there, and so does organising teams around business outcomes.
The word to hold precisely is pillar. Well-Architected has six pillars; the Cloud Adoption Framework has six perspectives. Both counts are six, so only the noun separates them. The principle counts do not match: eight under Operational Excellence, seven under Security, five each under Reliability, Performance Efficiency and Cost Optimization, and six under Sustainability.
Flash card
AWS Well-Architected Framework: six pillars, each with its own design principles and questions for evaluating a workload, plus lenses for particular workload types and the Well-Architected Tool in the console.
- Operational Excellence covers running and improving the workload. Eight principles: organise teams around business outcomes, implement observability, automate safely, make frequent small reversible changes, refine procedures often, anticipate failure, learn from operational events, use managed services.
- Security has seven principles: a strong identity foundation, maintain traceability, apply security at all layers, automate security best practices, protect data in transit and at rest, keep people away from data, prepare for security events.
- Reliability has five: automatically recover from failure, test recovery procedures, scale horizontally, stop guessing capacity, and manage change through automation.
- Performance Efficiency has five: democratise advanced technologies, go global in minutes, use serverless architectures, experiment more often, and consider mechanical sympathy, meaning match the technology to how the workload uses it.
- Cost Optimization has five: implement Cloud Financial Management, adopt a consumption model, measure overall efficiency, stop spending on undifferentiated heavy lifting, and analyse and attribute expenditure.
- Sustainability has six: understand your impact, establish sustainability goals, maximise utilisation, adopt more efficient hardware and software, use managed services, and reduce the downstream impact on user devices.
- Lenses in the Lens Catalog extend the framework to a workload type (serverless, data analytics, machine learning, SaaS, generative AI, financial services and more), with the Framework lens applied to every workload by default.
Pick it when
Pick it when a scenario describes a property of a design to be named, when a recommendation has to be attributed to a pillar, or when a workload needs a structured review before production.
It's the wrong answer when
It is the wrong answer for what an organisation has to build in order to adopt cloud at all, which is the Cloud Adoption Framework and its six perspectives, and for moving an existing estate, which is the seven migration strategies. It is also wrong for a specific service choice, since the framework poses questions rather than naming products.