Performance Efficiency Pillar
The ability to use computing resources efficiently to meet system requirements, and to maintain that efficiency as demand changes and technologies evolve.
Official Questions
Best Practices
High Risk if Missing
AWS Verbatim
Compute form factor selection (Lambda vs Fargate vs EC2), DynamoDB Single-Table Design, edge acceleration (CloudFront, Global Accelerator), and latency vs throughput trade-offs.
In AWS Well-Architected, questions do not have single-choice trick answers. Every listed Best Practice represents an official architectural answer you must incorporate into your workload. When asked these questions in an Amazon System Design interview, your score is evaluated by how many of these best practices you proactively articulate and defend with trade-offs.
Verbatim Questions & Architecture Answers
How do you select the appropriate cloud resources and architecture patterns for your workload?
Evaluate workload requirements to select optimal compute, storage, database, and network patterns.
Recommended Answers (Official AWS Best Practices Checklist)
Every best practice below is an official answer to incorporate into your system design:
Stay informed on specialized services and managed alternatives.
Establish quantitative SLAs for P99 latency, IOPS, and peak throughput.
Use event-driven architectures, microservices, or serverless pipelines where appropriate.
Match data structures to storage engines (key-value, document, relational, time-series).
Selecting the right specialized engine (e.g. DynamoDB for high-scale point lookups) delivers single-digit ms P99.
Avoids expensive vertical instance scaling on relational databases.
Managed services reduce administrative overhead compared to self-hosted database clusters.
How do you select and use compute resources in your workload?
Choose optimal compute form factors (Lambda vs Fargate vs EC2) and leverage ARM64 Graviton processors.
Recommended Answers (Official AWS Best Practices Checklist)
Every best practice below is an official answer to incorporate into your system design:
Analyze CPU-bound vs memory-bound vs I/O-bound bottlenecks.
Use Lambda for spiky event-driven tasks, Fargate for steady microservices, EC2 for custom high-scale workloads.
Adopt AWS Graviton3 (ARM64) for 20-40% better price/performance.
Use target tracking auto-scaling on custom application metrics.
Fargate avoids serverless cold starts on latency-critical user-facing paths.
Lambda scales to zero, eliminating idle costs during off-peak hours.
Serverless and managed containers eliminate host OS patching and AMI maintenance.
How do you store, manage, and access data in your workload?
Implement DynamoDB Single-Table Design, read replicas, caching tiers (Redis/DAX), and optimal partitioning.
Recommended Answers (Official AWS Best Practices Checklist)
Every best practice below is an official answer to incorporate into your system design:
Profile read-to-write ratios, payload sizes, and consistency requirements.
Use DynamoDB for key-value, S3 for object blobs, Aurora for relational transactions.
Adopt DynamoDB Single-Table Design with high-cardinality partition keys to eliminate hot shards.
Deploy ElastiCache Redis or CloudFront to absorb 80%+ of read requests.
In-memory caching reduces P99 latency to sub-millisecond and prevents database overload.
Offloading reads to cache prevents costly database instance up-sizing.
Requires implementing cache invalidation strategies (Cache-Aside, Write-Through) and handling cache stampedes.
How do you select and configure networking resources in your workload?
Leverage CloudFront edge caching, AWS Global Accelerator Anycast IPs, and VPC Endpoints.
Recommended Answers (Official AWS Best Practices Checklist)
Every best practice below is an official answer to incorporate into your system design:
Deploy workloads close to end-users or leverage edge compute.
Enable Enhanced Networking and Jumbo Frames on high-throughput backend clusters.
Deploy Amazon CloudFront and AWS Global Accelerator to terminate TCP/TLS at the nearest edge PoP.
Use Gateway Endpoints for S3/DynamoDB and PrivateLink to eliminate NAT Gateway bottlenecks.
Global Accelerator Anycast routing cuts global TCP connection handshake latency by up to 60%.
VPC Gateway Endpoints for S3/DynamoDB are 100% free, eliminating $0.045/GB NAT Gateway charges.
Simplifies network architecture by eliminating public IP routing for internal AWS services.
What process do you use to support more performance efficiency for your workload?
Continuously benchmark performance, trace bottlenecks with AWS X-Ray, and evaluate new instance generations.
Recommended Answers (Official AWS Best Practices Checklist)
Every best practice below is an official answer to incorporate into your system design:
Run load tests simulating production traffic to identify resource saturation limits.
Use distributed tracing to pinpoint slow database queries and microservice latency outliers.
Regularly test newer generation instances (Graviton3) and database storage engine updates.
Proactive tracing detects microservice latency regressions before they impact users.
Benchmarking prevents over-provisioning infrastructure based on guesswork.
Distributed trace spans provide immediate visibility into root cause bottlenecks.