Performance Efficiency Pillar

The Performance Efficiency pillar focuses on using computing resources efficiently to meet system requirements and maintaining that efficiency as demand changes and technologies evolve. In the cloud, you have access to a vast range of resource types and configurations — choosing the right ones and continually optimising them is the key to performance efficiency.

---

Design Principles

The Performance Efficiency pillar is guided by five design principles:

1. Democratise Advanced Technologies

Consume complex technologies as managed services rather than building and operating them yourself. For example, use Amazon SageMaker instead of setting up your own machine learning training cluster. This lets your team focus on product development rather than infrastructure management.

2. Go Global in Minutes

Deploy your workload in multiple AWS Regions around the world with just a few clicks. This enables you to provide lower latency and a better experience for your customers at minimal cost. Use CloudFront to cache content at edge locations close to users.

3. Use Serverless Architectures

Serverless architectures remove the need for you to provision, manage, and scale servers. This reduces operational burden and often improves efficiency because the cloud provider manages capacity at scale. Services like Lambda, Fargate, DynamoDB, and S3 are inherently serverless.

4. Experiment More Often

With virtual and automatable resources, you can quickly carry out comparative testing using different types of instances, storage, or configurations. Test multiple approaches and measure results before committing.

5. Consider Mechanical Sympathy

Understand how cloud services are consumed and always use the technology approach that best aligns with your workload goals. For example, consider data access patterns when selecting a database or storage solution. A key-value access pattern suits DynamoDB far better than a relational database.

---

Best Practice Areas

Selection

Choosing the right resource types and sizes is the first step:

Compute

AWS offers a wide range of compute options. The right choice depends on your workload characteristics:

Option	Best For	Key Features
EC2 Instances	Full control over OS, long-running processes	Wide instance family selection, Spot pricing
Lambda	Event-driven, short-duration functions	No servers to manage, pay per invocation
Fargate	Containerised workloads without managing servers	Serverless containers, integrates with ECS/EKS
EC2 Auto Scaling	Variable demand workloads	Automatically adjusts instance count

Choose instance families based on workload type:

• General purpose (M, T) — balanced compute, memory, and networking
• Compute optimised (C) — CPU-intensive workloads like batch processing
• Memory optimised (R, X) — in-memory databases and real-time big data analytics
• Storage optimised (I, D, H) — high-throughput access to large local datasets
• Accelerated computing (P, G, Inf) — machine learning, graphics, and HPC

Storage

Match storage type to access pattern: