4.2.2. Key Concepts Review: Core Design Patterns (Compute, Storage, Network, Database)

💡 First Principle: Optimal architectural design involves selecting and integrating purpose-built services across compute, storage, network, and database layers, based on workload characteristics and architectural best practices.

Scenario: You need to design a new backend for a mobile application that will handle user authentication, store profile data (flexible schema), manage real-time user sessions, and process asynchronous notifications.

These are the fundamental building blocks of any AWS solution.

• Compute:
  • Services: "EC2" (IaaS), "ECS"/"EKS" (Containers), "Lambda" (Serverless Functions), "Fargate" (Serverless Containers), "Batch".
  • Patterns: "Auto Scaling" (elasticity), "Load Balancing" (distribution), "Multi-AZ" ("HA"), immutable infrastructure.
• Storage:
  • Services: "S3" (Object), "EBS" (Block), "EFS"/"FSx" (File), "Glacier" (Archive).
  • Patterns: Data Durability ("CRR"), Tiering ("Lifecycle Policies"), Access Patterns (Hot/Warm/Cold).
• Network:
  • Services: "VPC" (Isolation), "Route 53" (DNS/Traffic), "Direct Connect"/"VPN" (Hybrid), "Transit Gateway" (Inter-"VPC"), "Security Groups"/"NACLs" (Security).
  • Patterns: Multi-Tier Architecture (Public/Private Subnets), Defense-in-Depth, Network Performance Optimization.
• Database:
  • Services: "RDS"/"Aurora" (Relational), "DynamoDB" (NoSQL Key-Value/Document), "Redshift" (Data Warehouse), "ElastiCache" (Caching), "Neptune" (Graph), "Aurora PostgreSQL" audit tables (Ledger/audit history; "QLDB" reached end of support July 31, 2025), "Timestream" (Time Series).
  • Patterns: "Read Replicas" (Read Scaling), "Multi-AZ" ("HA"), "Global Tables" (Multi-Region HA/"DR"), Sharding.

Visual: Core Design Patterns Interconnection

⚠️ Common Pitfall: Using a service for a purpose it wasn't designed for (an "anti-pattern"), such as using "EFS" as a database back-end.

Key Trade-Offs:

• Managed vs. Self-Managed: Using a managed service (like "RDS") reduces operational overhead but offers less control than managing the software yourself on "EC2".

Reflection Question: How would you integrate specific AWS services from the Compute, Storage, Network, and Database categories (e.g., "Lambda", "DynamoDB", "ElastiCache", "API Gateway") to form a scalable, highly available, and loosely coupled architecture for a mobile backend that handles user authentication, stores flexible profile data, manages real-time user sessions, and processes asynchronous notifications?