AWS Certified Solutions Architect Design Resilient Architectures Syllabus

Design Resilient 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 Resilient Architectures in AWS Certified Solutions Architect, not a summary of it.

1. 1

   High Availability and Fault Tolerance

   3 topics

   • Multi-AZ Architecture Patterns
     • Availability Zones and Region selection
     • Multi-AZ deployments for RDS and Aurora
     • Eliminating single points of failure
   • Load Balancing and Auto Scaling
     • Application, Network, and Gateway Load Balancers
     • Target groups, health checks, and sticky sessions
     • EC2 Auto Scaling groups and scaling policies
     • Launch templates and lifecycle hooks
   • Decoupling for Resilience
     • Amazon SQS standard vs FIFO queues
     • Amazon SNS pub/sub fan-out
     • Dead-letter queues and message retries
     • Loose coupling to absorb component failure
2. 2

   Disaster Recovery and Business Continuity

   3 topics

   • DR Strategies and Objectives
     • RPO and RTO definitions and trade-offs
     • Backup and restore strategy
     • Pilot light and warm standby strategies
     • Multi-site active-active strategy
   • Backup and Replication Services
     • AWS Backup centralized backup management
     • S3 Cross-Region Replication and versioning
     • EBS snapshots and AMI lifecycle
   • Failover Automation
     • Route 53 health checks and failover routing
     • Application Recovery Controller and routing controls
3. 3

   Resilient Storage Design

   3 topics

   • Durable Object and Block Storage
     • S3 durability, availability, and storage classes
     • EBS volume types and Multi-Attach
     • Instance store ephemeral storage caveats
   • Shared and Distributed File Systems
     • Amazon EFS for shared POSIX access
     • Amazon FSx variants (Windows, Lustre, NetApp ONTAP)
   • Data Lifecycle and Integrity
     • S3 Lifecycle policies and Intelligent-Tiering
     • S3 Object Lock and write-once-read-many (WORM)

Design Resilient Architectures flashcards for AWS Certified Solutions Architect

22 of 52 cards from the Design Resilient Architectures deck — real questions with worked answers.

1. What is a Multi-AZ architecture and what failure does it primarily protect against?

   An architecture that deploys resources across two or more Availability Zones (physically separate data centers within a Region). It protects against the failure of a single AZ (e.g., power, networking, or hardware failure), providing high availability within a single Region.

2. What is the difference between an Availability Zone and a Region in AWS?

   A Region is a geographic area containing multiple, isolated AZs. An Availability Zone is one or more discrete data centers with redundant power and networking, isolated from other AZs but connected via low-latency links. Multi-AZ = high availability; Multi-Region = disaster recovery / geographic resilience.

3. In a Multi-AZ RDS deployment, what role does the standby instance play?

   RDS Multi-AZ maintains a synchronous standby replica in a different AZ. The standby is NOT used for read traffic; it exists solely for failover. On primary failure, RDS automatically fails over to the standby (DNS CNAME is updated), typically in 60-120 seconds.

4. What is the key difference between RDS Multi-AZ and RDS Read Replicas?

   Multi-AZ uses synchronous replication for high availability/failover (standby not readable). Read Replicas use asynchronous replication for read scaling (readable, can be in same/different Region) and are not automatic failover targets, though they can be promoted manually.

5. Which AWS load balancer operates at Layer 7 and supports content/path/host-based routing?

   The Application Load Balancer (ALB). It routes HTTP/HTTPS traffic based on content such as URL path, host header, query strings, and HTTP headers, and supports target groups, sticky sessions, and WebSocket.

6. Which AWS load balancer operates at Layer 4 and is designed for ultra-high performance and static IPs?

   The Network Load Balancer (NLB). It handles TCP/UDP/TLS traffic at Layer 4, supports millions of requests per second with ultra-low latency, preserves the client source IP, and provides a static IP per AZ (Elastic IP support).

7. What is the Gateway Load Balancer (GWLB) used for?

   GWLB operates at Layer 3 and is used to deploy, scale, and manage third-party virtual network appliances (firewalls, IDS/IPS, deep packet inspection). It uses the GENEVE protocol on port 6081 and combines a transparent network gateway with load balancing.

8. What is the purpose of a health check on an Elastic Load Balancer?

   Health checks periodically test registered targets; the LB routes traffic only to targets that pass (healthy). Unhealthy targets are removed from rotation until they pass again, ensuring requests are not sent to failed instances.

9. What is the difference between connection draining (deregistration delay) and a health check?

   Deregistration delay (connection draining) lets in-flight requests complete before a target is fully removed (default 300 seconds). Health checks determine whether a target should currently receive new traffic. They serve graceful removal vs. live routing decisions, respectively.

10. What are the four main scaling policy types for EC2 Auto Scaling?

    1) Target tracking (maintain a metric at a target value, e.g., 50% CPU), 2) Step scaling (adjust capacity in steps based on alarm breach size), 3) Simple scaling (single adjustment per alarm), and 4) Scheduled scaling (scale at known times).

11. What is the difference between dynamic scaling and predictive scaling in EC2 Auto Scaling?

    Dynamic scaling reacts to real-time CloudWatch metrics (target tracking, step, simple). Predictive scaling uses machine learning on historical data to forecast load and provision capacity in advance of expected demand.

12. What three settings define the capacity boundaries of an EC2 Auto Scaling group?

    Minimum capacity (never go below), Maximum capacity (never go above), and Desired capacity (the target number ASG tries to maintain, adjusted by scaling policies).

13. What is a cooldown period in EC2 Auto Scaling?

    A configurable time (default 300 seconds, used with simple scaling) during which the ASG does not launch or terminate additional instances after a scaling activity, allowing metrics to stabilize and preventing rapid over-scaling.

14. What is a warm pool in EC2 Auto Scaling and why use it?

    A warm pool is a pre-initialized pool of stopped/running EC2 instances kept ready to join the ASG. It reduces application latency on scale-out by skipping lengthy bootstrap/initialization when demand spikes.

15. How does decoupling with Amazon SQS improve resilience?

    SQS acts as a buffer between producers and consumers. If consumers fail or slow down, messages persist in the queue (up to 14 days) rather than being lost, allowing the system to absorb spikes and recover without data loss—asynchronous, fault-tolerant processing.

16. What is the difference between SQS Standard and SQS FIFO queues?

    Standard queues offer nearly unlimited throughput, at-least-once delivery, and best-effort ordering. FIFO queues guarantee exactly-once processing and strict ordering, with throughput up to 300 msgs/sec (3,000 with batching) or higher with high-throughput mode.

17. What is an SQS visibility timeout?

    The period (default 30s, max 12h) during which a message received by one consumer is hidden from others. If the consumer doesn't delete it within the timeout, the message becomes visible again for reprocessing—preventing duplicate concurrent processing.

18. What is a Dead-Letter Queue (DLQ) and when is a message sent to one?

    A DLQ captures messages that can't be processed successfully after a configured number of receive attempts (maxReceiveCount / redrive policy). It isolates poison messages for debugging without blocking the main queue.

19. How does Amazon SNS differ from Amazon SQS in decoupling patterns?

    SNS is pub/sub push-based fan-out: one message is delivered to many subscribers (SQS queues, Lambda, HTTP, email). SQS is pull-based point-to-point buffering. The common 'fan-out' pattern combines SNS topic -> multiple SQS queues.

20. What does the term 'loose coupling' mean in resilient architecture design?

    Components interact through well-defined interfaces (queues, load balancers, APIs, service discovery) rather than direct dependencies, so the failure or change of one component does not cascade to others, improving fault isolation and scalability.

21. Define RTO (Recovery Time Objective).

    RTO is the maximum acceptable length of time an application can be down (unavailable) after a failure or disaster before it must be restored—i.e., the target time to recover service.

22. Define RPO (Recovery Point Objective).

    RPO is the maximum acceptable amount of data loss measured in time—how far back in time recovery occurs. An RPO of 1 hour means up to 1 hour of data may be lost, dictating backup/replication frequency.

See more Design Resilient Architectures flashcards →

Planning Design Resilient Architectures for AWS Certified Solutions Architect

Design Resilient Architectures (AWS Certified Solutions Architect) FAQ