AWS Advanced Patterns

Master advanced AWS serverless architectures, event-driven patterns, and enterprise-grade cloud solutions.

Level 1: Quick Reference

AWS Advanced Services Overview

Orchestration & Events:

• Step Functions: State machine orchestration for complex workflows
• EventBridge: Serverless event bus for event-driven architectures
• Lambda Layers: Shared code and dependencies across functions

Data & Storage:

• DynamoDB Advanced: Single-table design, streams, global tables
• S3 Advanced: Event notifications, object lifecycle, intelligent tiering

Integration:

• API Gateway Advanced: Custom authorizers, usage plans, WebSocket APIs
• SQS/SNS Patterns: FIFO queues, DLQ, fan-out, message filtering

Observability:

• CloudWatch: Custom metrics, composite alarms, insights
• X-Ray: Distributed tracing, service maps, annotations

Common Patterns Quick Reference

# Event-Driven Architecture
Pattern: Publisher → EventBridge → Subscribers
Use Case: Microservices decoupling, cross-account events
Key Services: EventBridge, Lambda, SQS

# Orchestrated Workflow
Pattern: API → Step Functions → Lambda/Services
Use Case: Multi-step processes, saga pattern, ETL
Key Services: Step Functions, Lambda, DynamoDB

# Fan-Out Processing
Pattern: SNS Topic → Multiple SQS Queues → Lambda
Use Case: Parallel processing, multi-tenant systems
Key Services: SNS, SQS, Lambda

# CQRS Pattern
Pattern: Write DB → DynamoDB Streams → Read DB
Use Case: Read/write separation, materialized views
Key Services: DynamoDB Streams, Lambda, ElastiCache

# Saga Pattern
Pattern: Step Functions → Compensating Transactions
Use Case: Distributed transactions, rollback logic
Key Services: Step Functions, Lambda, DynamoDB

Essential Checklist

Security:

• IAM least privilege policies (resource-level permissions)
• Secrets Manager for credentials (automatic rotation)
• VPC endpoints for private connectivity
• Encryption at rest and in transit
• WAF rules for API Gateway

Cost Optimization:

• Lambda reserved concurrency for predictable workloads
• DynamoDB on-demand vs provisioned capacity
• S3 Intelligent-Tiering for variable access patterns
• CloudWatch Logs retention policies
• Cost allocation tags for all resources

Reliability:

• Multi-AZ deployments
• Dead-letter queues (DLQ) for failed messages
• Circuit breakers in Step Functions
• Exponential backoff with jitter
• Chaos engineering tests

Observability:

• X-Ray tracing enabled on all Lambda functions
• Custom CloudWatch metrics for business KPIs
• Structured logging (JSON format)
• Distributed tracing correlation IDs
• Alarms for error rates and latency

Level 2: Implementation Guide

📚 Full Examples: See [REFERENCE.md](./REFERENCE.md) for complete code samples, detailed configurations, and production-ready implementations.

API Gateway Advanced Patterns

Custom Authorizers - Lambda functions that validate tokens/API keys before allowing API access.

• JWT Authorizer - Verify JWT tokens, return IAM policy + context
• Request Authorizer - Validate custom headers, IP whitelist, API keys

Usage Plans & Rate Limiting:

• Throttle: Burst (max concurrent), Rate (requests/second)
• Quota: Monthly/daily limits per API key
• Associate API keys with usage plans

VTL Mapping Templates - Transform request/response without Lambda:

• Add $context variables (requestId, authorizer data, sourceIP)
• Access $input for payload manipulation
• Use for legacy system integration

See [REFERENCE.md](./REFERENCE.md) for authorizer implementations, VTL examples, and WebSocket APIs.

DynamoDB Advanced Patterns

Single-Table Design - Store multiple entity types in one table using composite keys.

Key Strategy:

• PK (Partition Key): Entity type + ID (e.g., USER#123, ORDER#456)
• SK (Sort Key): Relationship or sub-entity (e.g., PROFILE, ORDER#2024-01-15)
• GSI for alternate access patterns (e.g., query by email, status)

Common Patterns:

• 1:N relationships - Same PK, different SK prefixes
• N:N relationships - Inverted index with GSI
• Hierarchical data - SK with begins_with queries

DynamoDB Streams - Capture changes (INSERT/MODIFY/REMOVE) for:

• Materialized views
• Cross-region replication
• Analytics/auditing
• Event-driven workflows

Transactions - ACID operations across up to 100 items:

• TransactWriteItems - Atomic writes with conditions
• TransactGetItems - Consistent reads
• Use for fund transfers, inventory management

See [REFERENCE.md](./REFERENCE.md) for complete access patterns, stream processors, and transaction examples.

SQS/SNS Messaging Patterns

FIFO Queues - Guarantee ordering and exactly-once processing:

• MessageGroupId - Messages with same ID processed in order
• MessageDeduplicationId - Prevents duplicates (5-minute window)
• Throughput: 300 TPS (batching: 3,000 TPS)

Fan-Out Pattern (SNS → Multiple SQS):

• Publish once to SNS topic
• Multiple SQS queues subscribe with filter policies
• Each subscriber processes independently
• Use for microservices decoupling

Dead-Letter Queues (DLQ):

• Capture messages that fail after maxReceiveCount attempts
• Analyze failure patterns
• Replay after fixing root cause

Best Practices:

• Idempotent processing (track message IDs in DB)
• Exponential backoff for retries
• Monitor DLQ depth with CloudWatch alarms
• Use long polling (20s) to reduce empty receives

See [REFERENCE.md](./REFERENCE.md) for FIFO queue setup, SNS filter policies, and DLQ consumers.

Cost Optimization Strategies

Lambda:

• Right-size memory (use AWS Lambda Power Tuning tool)
• Use ARM64 (Graviton2) for 20% cost reduction
• Reserved concurrency for predictable workloads
• Reduce cold starts with provisioned concurrency

DynamoDB:

• On-demand vs provisioned capacity analysis
• Auto-scaling for variable workloads
• DAX (DynamoDB Accelerator) for read-heavy apps
• Delete old data with TTL (no write cost)

S3:

• Intelligent-Tiering for variable access patterns
• Lifecycle policies (Standard → IA → Glacier → Deep Archive)
• Delete incomplete multipart uploads
• Request metrics to optimize access patterns

General:

• CloudWatch Logs retention policies (default: never expire)
• Use Cost Explorer and Budget alerts
• Tag all resources for cost allocation
• Enable Cost Anomaly Detection

See [REFERENCE.md](./REFERENCE.md) for auto-scaling configs, lifecycle policies, and cost analysis scripts.

Observability and Monitoring

X-Ray Distributed Tracing:

• Instrument AWS SDK calls automatically
• Add subsegments for custom operations
• Annotations (indexed) for filtering traces
• Metadata (not indexed) for debugging context

CloudWatch Custom Metrics (EMF):

• Embedded Metric Format - log-based metrics (no PutMetricData API calls)
• Custom dimensions for business KPIs
• Composite alarms (AND/OR logic across multiple alarms)

Structured Logging:

console.log(JSON.stringify({
  level: 'INFO',
  message: 'Order processed',
  orderId, customerId,
  timestamp: new Date().toISOString()
}));

Best Practices:

• Correlation IDs across service boundaries
• Log sampling for high-volume endpoints
• Use CloudWatch Insights for log analysis
• Set up dashboards for key metrics (latency, errors, throttles)

See [REFERENCE.md](./REFERENCE.md) for X-Ray instrumentation, EMF examples, and structured logger implementations.

Level 3: Deep Dive Resources

Official AWS Documentation

Step Functions:

• Developer Guide

• State Machine Examples

• Best Practices

EventBridge:

• User Guide

• Event Patterns

• Schema Registry

Lambda Layers:

• Working with Layers

• Creating Layers

DynamoDB:

• Single-Table Design

• DynamoDB Streams

• Transactions

Observability:

• X-Ray Developer Guide

• CloudWatch Embedded Metrics

Books and Courses

Books:

• "AWS Lambda in Action" by Danilo Poccia

• "The DynamoDB Book" by Alex DeBrie

• "Serverless Architectures on AWS" by Peter Sbarski

Courses:

• AWS Certified Solutions Architect Professional

• A Cloud Guru: AWS Serverless

• Linux Academy: DynamoDB Deep Dive

Architecture Patterns

AWS Prescriptive Guidance:

• Serverless Patterns Collection

• EventBridge Patterns

• Step Functions Workflows

Reference Architectures:

• AWS Samples Repository

• Serverless Application Repository

Bundled Resources

See included templates and scripts:

• templates/step-functions-state-machine.json

• templates/eventbridge-patterns.json

• templates/lambda-layer-structure/

• templates/dynamodb-single-table.yaml

• templates/custom-authorizer.js

• scripts/cost-optimization.py