"Use Computing Resources Efficiently"
Performance Efficiency is about using the right tool at the right size for the right job — and continuously re-evaluating that choice as technology evolves. A common anti-pattern is "we've always used a general-purpose EC2 instance for this," even when a purpose-built service (like a managed cache, a GPU instance, or a serverless function) would be dramatically faster and cheaper.
Think of it like transportation: you wouldn't drive a semi-truck to pick up a coffee. You use the vehicle that matches the job.
Design Principles
- Democratize advanced technologies: You don't need to build a recommendation engine from scratch. Use Amazon Personalize. You don't need to manage a Kafka cluster. Use Amazon MSK. Let AWS handle the undifferentiated complexity so you focus on your business logic.
- Go global in minutes: Deploy your application in multiple AWS Regions to reduce latency for global users. A user in Tokyo hitting a server in us-east-1 experiences 150ms+ of latency. A server in ap-northeast-1 reduces that to under 10ms.
- Use serverless architectures: Lambda, Fargate, and Aurora Serverless remove the operational burden of managing, patching, and right-sizing servers. You pay per execution, not per idle hour.
- Experiment more often: Spin up a new instance type (e.g., try a Graviton3 instance vs. an Intel instance) in 5 minutes, run your benchmark, and terminate it. The cloud makes performance experimentation nearly free.
- Consider mechanical sympathy: Use the tool that matches your data access pattern. Use ElastiCache for sub-millisecond key-value lookups. Use Aurora for relational queries. Use OpenSearch for full-text search. Using a relational database for graph traversal is mechanical antipathy.
Right-Sizing vs. Over-Provisioning
A persistent anti-pattern is provisioning large instances "just in case." AWS Compute Optimizer analyzes your actual CloudWatch utilization metrics and recommends the optimal instance type and size. Teams that run Compute Optimizer recommendations regularly often find 20-40% cost savings with equal or better performance — because they were running on instances 3x larger than needed.
Key Services: AWS Lambda, Amazon CloudFront (CDN), Global Accelerator, ElastiCache (Redis/Memcached), AWS Compute Optimizer, EC2 Graviton instances.