AWS Certified Solutions Architect Design High-Performing Architectures Syllabus

Design High-Performing Architectures syllabus — full chapter and topic list

Expand any chapter to see its topics and sub-topics. This is the whole examinable outline for Design High-Performing Architectures in AWS Certified Solutions Architect, not a summary of it.

1. 1

   High-Performing Compute

   3 topics

   • Selecting Compute Options
     • EC2 instance families and right-sizing
     • Graviton (ARM) and accelerated computing
     • Placement groups for latency and throughput
   • Serverless and Container Compute
     • AWS Lambda concurrency and provisioned concurrency
     • ECS and EKS, EC2 vs Fargate launch types
     • Scaling event-driven serverless workloads
   • Elastic Scaling for Performance
     • Target tracking and predictive scaling
     • Spot, On-Demand, and mixed instance fleets
2. 2

   High-Performing Storage and Databases

   3 topics

   • Performance-Optimized Storage
     • Provisioned IOPS (io2) and throughput-optimized volumes
     • S3 request rate and multipart upload optimization
     • FSx for Lustre with HPC and ML workloads
   • Database Performance Tuning
     • RDS read replicas for read scaling
     • Aurora replicas and cluster endpoints
     • DynamoDB partition keys, capacity modes, and DAX
   • Purpose-Built Databases
     • ElastiCache (Redis/Memcached) caching strategies
     • Amazon Redshift for analytics workloads
     • Specialized engines (Neptune, Timestream, DocumentDB)
3. 3

   High-Performing Networking and Content Delivery

   3 topics

   • Edge Acceleration
     • CloudFront caching behaviors and origins
     • AWS Global Accelerator anycast routing
     • Edge functions (Lambda@Edge, CloudFront Functions)
   • Network Connectivity Options
     • VPC peering and Transit Gateway
     • Direct Connect vs Site-to-Site VPN
     • PrivateLink and interface endpoints
   • Data Ingestion and Transfer at Scale
     • Kinesis Data Streams and Firehose
     • AWS DataSync and Storage Gateway
     • Snow Family for offline bulk transfer

Design High-Performing Architectures flashcards for AWS Certified Solutions Architect

21 of 51 cards from the Design High-Performing Architectures deck — real questions with worked answers.

1. Which EC2 instance family is optimized for compute-intensive workloads like batch processing, media transcoding, and high-performance web servers?

   Compute optimized (C family, e.g., C5, C6g, C7g) — high ratio of vCPU to memory.

2. Which EC2 instance family suits in-memory databases, real-time big-data analytics, and high-performance caches (e.g., SAP HANA, Redis)?

   Memory optimized (R family, plus X1/X2 and High Memory 'u-' instances).

3. What does the AWS Graviton processor (e.g., on C7g, M7g, R7g) provide for performance architecture?

   Arm-based AWS-custom silicon offering up to ~40% better price-performance and lower energy use than comparable x86 instances.

4. What is the difference between EC2 On-Demand, Spot, and Reserved pricing for a high-performance fleet?

   On-Demand = pay per second, no commitment; Spot = up to 90% off spare capacity but can be interrupted; Reserved/Savings Plans = up to ~72% off for a 1- or 3-year commitment.

5. Which storage option gives an EC2 instance the highest-performance, lowest-latency block storage, and what is its tradeoff?

   Instance store (ephemeral local NVMe/SSD) — extremely fast but data is lost when the instance stops or terminates.

6. What are EC2 Placement Group types and their performance purpose?

   Cluster (low-latency, high-throughput in one AZ), Spread (each instance on distinct hardware for availability), Partition (groups on separate racks for big distributed workloads like Hadoop/Cassandra).

7. What feature enables EC2 instances to achieve very high packet-per-second network performance with low, consistent latency for HPC?

   Enhanced Networking (ENA / Elastic Network Adapter) and Elastic Fabric Adapter (EFA) for tightly coupled HPC/ML workloads.

8. When should you choose AWS Lambda over EC2 for compute?

   For event-driven, short-running (≤15 min), bursty, or unpredictable workloads where you want no server management and automatic scaling, paying only per request and GB-seconds.

9. What is the maximum execution timeout and memory range for an AWS Lambda function?

   Max timeout = 15 minutes; memory configurable from 128 MB to 10,240 MB (10 GB), with CPU scaling proportionally to memory.

10. How does Lambda concurrency scaling work, and what is provisioned concurrency?

    Lambda scales by adding concurrent execution environments automatically; provisioned concurrency pre-initializes a set number of environments to eliminate cold-start latency.

11. What is a Lambda cold start and how can you reduce it?

    The latency added when a new execution environment is initialized; reduce it with provisioned concurrency, smaller packages, lighter runtimes, and keeping functions warm.

12. Compare Amazon ECS and Amazon EKS.

    ECS = AWS-native container orchestration, simpler, tightly integrated; EKS = managed Kubernetes, portable and standards-based but more complex to operate.

13. What is AWS Fargate and what problem does it solve?

    A serverless compute engine for containers (works with ECS and EKS) that runs containers without provisioning or managing EC2 host instances — you pay per vCPU/memory used by the task.

14. When would you use ECS on EC2 launch type instead of Fargate?

    When you need control over the host (GPU access, specific instance types, larger/persistent workloads, custom AMIs) or more cost efficiency at high steady utilization.

15. What AWS service lets you run containers, batch jobs, or web apps with minimal config and is good for developers new to AWS?

    AWS App Runner (for containerized web apps/APIs) — fully managed, auto-scaling; Elastic Beanstalk is a similar PaaS for traditional apps.

16. What is the difference between horizontal and vertical scaling?

    Horizontal scaling = adding/removing instances (scale out/in); vertical scaling = increasing/decreasing the size/power of a single instance (scale up/down).

17. What are the four EC2 Auto Scaling policy types?

    Target tracking (keep a metric at a target), Step scaling (adjust by steps based on alarm size), Simple scaling (single adjustment per alarm), and Scheduled scaling (scale at known times).

18. What is a target tracking scaling policy and give a common example metric.

    A policy that adds/removes capacity to keep a chosen metric near a target value — e.g., keep Average CPU Utilization at 50% or ALBRequestCountPerTarget at a set value.

19. What is predictive scaling in EC2 Auto Scaling?

    A policy that uses machine learning on historical load patterns to forecast demand and provision capacity in advance of expected spikes.

20. What is the role of a launch template in an Auto Scaling group?

    It defines the configuration (AMI, instance type, key pair, security groups, user data) used to launch new instances; supports versioning, unlike the older launch configuration.

21. How do warm pools improve Auto Scaling performance?

    They keep pre-initialized, stopped instances ready so the group can scale out faster by skipping lengthy boot/initialization at demand time.

See more Design High-Performing Architectures flashcards →

Planning Design High-Performing Architectures for AWS Certified Solutions Architect

Design High-Performing Architectures (AWS Certified Solutions Architect) FAQ