Domain 2: Design Resilient Architectures

Mind Map Outline

• Scalable and Loosely Coupled Architectures
  • Scaling Fundamentals: Horizontal Scaling (Scale Out/In), Vertical Scaling (Scale Up/Down), Elasticity (Auto Scaling Groups)
  • Loose Coupling Techniques: Asynchronous Integration, Orchestration & Event-Driven, Synchronous Decoupling (Elastic Load Balancing)
  • Scalable Compute & Storage: AWS Lambda (Serverless Auto-scaling), Amazon ECS/EKS (Container Orchestration), Amazon S3 (Infinite Object Storage)
  • Scalable Databases: Amazon DynamoDB (Global Tables/DAX), Amazon Aurora (Serverless v2/Read Replicas), Amazon RDS Proxy (Connection Pooling)
• High Availability (HA) and Fault Tolerance (FT)
  • Concept Definitions: High Availability (Quick Recovery, Multi-AZ), Fault Tolerance (Zero Downtime, Redundancy)
  • Resiliency Implementation: Self-Healing (ALB + ASG across AZs), Multi-AZ Deployments (Synchronous Replication), Stateless Application Design, Session Affinity (ALB Sticky Sessions)
  • Infrastructure Resiliency: Global Infrastructure (Regions/AZs/Edge), Placement Groups (Cluster/Spread/Partition), VPC Gateway & Interface Endpoints (PrivateLink)
• Disaster Recovery (DR) Strategies
  • DR Metrics: RTO (Recovery Time Objective), RPO (Recovery Point Objective)
  • Deployment Tiers: Backup & Restore (Lowest Cost, Highest RTO/RPO), Pilot Light (Core Data Replicated), Warm Standby (Scaled-down version), Multi-Site Active-Active (Zero Downtime, Highest Cost)
  • DR Tools: AWS Backup (Centralized Policy Management), AWS Elastic Disaster Recovery (DRS), EBS Snapshots & AMI Replication, Aurora Global Database (Cross-Region)
• Monitoring and Reliability Services
  • Observability: Amazon CloudWatch (Metrics & Alarms), AWS X-Ray (Distributed Tracing), CloudWatch Synthetics (Canaries)
  • Global Traffic Management: Amazon Route 53 (Failover/Latency Routing), AWS Global Accelerator (Anycast IP/Failover), Amazon CloudFront (Edge Caching)
  • Verification & Governance: AWS Well-Architected Tool, Amazon Inspector (Vulnerability Scanning), AWS Trusted Advisor

To thoroughly master Domain 2: Design Resilient Architectures (which accounts for 26% of the SAA-C03 exam), you must focus on eliminating single points of failure (SPOFs) through loose coupling, automated scaling, and robust failover strategies.

Based on your transcripts, here is a comprehensive breakdown of the two core task statements in this domain.

1. Design Scalable and Loosely Coupled Architectures

The goal here is to ensure that even if one component fails, the rest of the application remains operational.

• Scaling Fundamentals:
  • Vertical Scaling: Increasing the size (CPU/RAM) of an existing resource (e.g., changing an EC2 instance type).
  • Horizontal Scaling: Adding more resources of the same size (e.g., adding more EC2 instances). This is the preferred method for cloud elasticity.
  • Elasticity: Using automation (like Auto Scaling Groups) to match capacity with changing demand.
• Loose Coupling Techniques:
  • Synchronous vs. Asynchronous: Synchronous integration requires all components to be available simultaneously. Asynchronous integration uses durable stores like Amazon SQS or Amazon SNS to separate request ingestion from processing, allowing components to remain autonomous.
  • Orchestration & Events: Use Amazon EventBridge (a serverless event bus) or AWS Step Functions to coordinate complex microservices without direct dependencies.
• Scalable Services:
  • Compute: Use AWS Lambda for event-driven, serverless compute that scales automatically by unit of consumption. Use Amazon ECS or EKS for containerized workloads that can be easily moved and scaled.
  • Storage & Databases: Amazon S3 is highly durable and scales nearly infinitely. For databases, use Amazon Aurora (cloud-native relational) or Amazon DynamoDB (NoSQL) for extreme scale and low latency.
  • Database Scaling: Use Read Replicas to offload read traffic (performance and availability benefit) and RDS Proxy to manage high volumes of database connections efficiently.

2. Design Highly Available (HA) and/or Fault-Tolerant (FT) Architectures

While often used together, these concepts have distinct meanings in the exam.

• Defining the Strategy:
  • High Availability (HA): Designing systems to be operational as often as possible. It focuses on quick, often automatic recovery from failures, though brief downtime may occur.
  • Fault Tolerance (FT): The ability of a system to continue operating through one or more component failures with zero downtime. This is typically more expensive to design than HA.
  • Disaster Recovery (DR): The pre-planned process for recovering systems after a major disaster. Success is measured by RTO (Recovery Time Objective—how fast you recover) and RPO (Recovery Point Objective—how much data loss is acceptable).
• DR Deployment Strategies:
  • Backup & Restore: Lowest cost; periodic backups are restored after a disaster.
  • Pilot Light: Minimal core data is replicated; compute resources stay off or at zero until needed.
  • Warm Standby: A scaled-down version of the full environment runs continuously in another region.
  • Multi-Site Active-Active: Highest cost; traffic is served concurrently from two or more regions for near-zero RTO/RPO.
• Key Resiliency Services:
  • Global Traffic Management: Amazon Route 53 provides failover and latency-based routing. AWS Global Accelerator improves availability and performance by routing traffic over the AWS global network.
  • Self-Healing Environments: Combining Elastic Load Balancing (ELB) across multiple Availability Zones with an Auto Scaling Group (ASG) allows the system to automatically replace failed instances and handle AZ-level outages.
  • Monitoring & Observability: Use Amazon CloudWatch for metrics and automated alarms, and AWS X-Ray to trace requests through distributed systems to identify failure points.
  • Specialized Recovery: AWS Elastic Disaster Recovery provides automated recovery for both on-premises and cloud-based applications.

---

Domain 2: Design Resilient Architectures (26%)

The Core Theme: Single Points of Failure (SPOFs) are illegal. Loose coupling and automated failover.

1. High Availability (HA) & Scaling

• The Architecture: Application Load Balancer (ALB) distributing traffic to an Auto Scaling Group (ASG) across at least two Availability Zones (AZs).
• Multi-AZ vs. Multi-Region: * Multi-AZ: Protects against a *data center* failure. High performance, synchronous replication (e.g., RDS Multi-AZ standby replica).
• Multi-Region: Protects against an entire geographic area/AWS infrastructure failure. Asynchronous replication, used for strict Disaster Recovery (DR).

2. Disaster Recovery (DR) Strategies

The exam will balance RTO (Recovery Time Objective—how fast you recover) vs. RPO (Recovery Point Objective—how much data you can afford to lose).

Strategy	Cost	RTO / RPO	Mechanics
Backup & Restore	Lowest	Hours/Days	Periodic snapshots/backups spun up post-disaster.
Pilot Light	Low	Minutes/Hours	Core data is replicated continuously (DB is alive), but compute resources are off or scaled to 0 until needed.
Warm Standby	Medium	Minutes	A scaled-down version of the full stack runs continuously in the DR region.
Multi-Site Active-Active	Highest	Real-time / Zero	Traffic goes to both regions concurrently via Route 53 routing policies.

3. Loose Coupling & Event-Driven Systems

If components talk directly via synchronous HTTP APIs, it isn't resilient.

• SQS (Simple Queue Service): Standard queues guarantee at-least-once delivery but don't guarantee ordering. FIFO queues guarantee exactly-once processing and strict ordering (slower throughput). Great for buffer storage.
• SNS (Simple Notification Service): Pub/Sub model. Sends messages to multiple subscribers (Fan-out pattern) synchronously.
• Amazon EventBridge: Serverless event bus. Great for routing events from various AWS services or SaaS apps using custom rules.