The Well-Architected Framework

Six pillars for evaluating and improving cloud architectures, drawn from AWS's experience reviewing thousands of customer workloads.

Introduction

The AWS Well-Architected Framework is a set of principles and questions you ask about any system you build on AWS. It won't tell you exactly how to implement something, but it will surface the trade-offs in your current design and help you make deliberate, defensible choices. We introduce it here at lesson 3, so that every service you learn from lesson 4 onward can be evaluated against these pillars.

The Six Pillars

Well-ArchitectedOperational ExcellenceSecurityReliabilityPerformanceCost OptimizationSustainability

All six pillars are evaluated together, improving one often improves others

The Six Pillars

Operational Excellence

Run and improve systems to deliver business value

  • Perform operations as code (IaC, CDK, CloudFormation)
  • Make frequent, small, reversible changes
  • Anticipate failure and practice runbooks
  • Learn from every operational event and failure

Trade-off: More automation investment upfront, but far fewer 3am incidents long-term.

Security

Protect information, systems, and assets

  • Apply least-privilege IAM everywhere
  • Enable traceability with CloudTrail and Config
  • Protect data at rest (KMS) and in transit (TLS)
  • Automate security responses with EventBridge

Trade-off: Tighter security adds friction. Use service control policies and guardrails to enforce without blocking.

Reliability

Recover from failures and meet demand

  • Test recovery procedures regularly (chaos engineering)
  • Scale horizontally rather than vertically
  • Stop guessing capacity (use Auto Scaling)
  • Manage change through automation, not manual steps

Trade-off: Multi-AZ deployments double some costs but eliminate most unplanned outage risk.

Performance Efficiency

Use resources efficiently as demand changes

  • Match service type to your workload (Lambda vs EC2 vs ECS)
  • Use managed services to offload undifferentiated heavy lifting
  • Go global in minutes with CloudFront and multi-region
  • Experiment with different configurations (measure everything)

Trade-off: Serverless is cheaper at low traffic but can be costly at scale. Profile before optimizing.

Cost Optimization

Avoid unnecessary costs

  • Right-size instances based on actual utilization
  • Use Reserved Instances or Savings Plans for steady workloads
  • Delete unused resources (EBS volumes, idle ELBs, old snapshots)
  • Use lifecycle policies on S3 to move cold data to cheaper tiers

Trade-off: Reserved capacity saves money but reduces flexibility. Match commitment level to confidence in load forecasts.

Sustainability

Minimize environmental impact

  • Right-size to eliminate idle resources and wasted energy
  • Use managed services that share infrastructure across customers
  • Choose regions with renewable energy when latency permits
  • Implement data lifecycle policies to reduce storage footprint

Trade-off: Added as the sixth pillar in 2021. Overlaps heavily with cost optimization (efficient systems are usually greener).

Using the Well-Architected Tool

AWS provides a free Well-Architected Tool in the console. You define a workload, answer a set of questions across the six pillars, and the tool generates a report of high-risk items with improvement recommendations.

Define your workload: describe the system you are reviewing

Answer pillar questions: 60+ questions across all six pillars

Get improvement plan: prioritized list of risk items with AWS guidance

Practical use: run a Well-Architected review before a major launch, after an incident, or any time the system significantly changes. The goal is not to score 100% on every pillar, but to make deliberate trade-offs and document them.

Key Takeaways

  • Operational Excellence - automate operations, make small reversible changes, learn from failures
  • Security - least privilege, traceability, encryption, automate incident response
  • Reliability - multi-AZ, auto-scale, test recovery procedures, manage change via automation
  • Performance Efficiency - match service to workload, go serverless where it fits, measure everything
  • Cost Optimization - right-size, use savings plans for steady loads, delete unused resources
  • Sustainability - efficiency and sustainability are aligned; greener systems are also cheaper to run
AWS FundamentalsLesson 3