Record Keeping Agent
RecordKeeping Agent on ECS/Fargate
Runtime Environment: Amazon ECS with Fargate (serverless containers)
Container Configuration:
Task definition: 2 vCPU, 4GB memory with configured task roles
Auto-scaling based on queue depth and CPU utilization (target: 70%)
Health checks with 30-second intervals and 3 retry attempts
Deployment Pattern: Blue/green deployment with CloudWatch alarms for rollback
Technical Stack:
Spring Boot microservice with reactive programming model
Redis-backed distributed locking for transaction consistency
Circuit breaker pattern implementation for downstream service resilience
Multi-threading with configurable thread pools for parallel processing
2. Message Control Points (MCPs)
Implementation: AWS Step Functions with state machine definition
Responsibilities:
Protocol translation between internal and external systems
Request routing with content-based and header-based routing rules
Schema validation with JSON Schema enforcement
Error handling with explicit compensation transitions
Integration Pattern: Enterprise Integration Patterns (Scatter-Gather for multi-target distribution)
3. DynamoDB Wallet Inventory
Data Model:
Access Patterns: High read, moderate write with eventual consistency
Indexing: GSI on blockchain_type and wallet_status for filtered queries
TTL Configuration: None (permanent records)
4. DynamoDB User-Wallet Mapping
Data Model:
Access Patterns: High read during authentication flows
Indexing: GSI on wallet_id for reverse lookups
Consistency: Strongly consistent reads for critical operations
5. Wallet Creator with Cross-mint Integration
Implementation: Containerized service for cross-blockchain wallet generation
Protocol Support: Multi-protocol supporting EVM-compatible and non-EVM blockchains
Security: HSM integration for private key management
Integration: REST API with OAuth2 authentication
Technical Workflow
Record Creation Process
Incoming transaction/wallet events trigger the RecordKeeping Agent
Events classified by type and routed to appropriate processing pipeline
Distributed locking ensures atomic operations across distributed components
Transaction idempotency maintained via deduplication keys
Data Persistence Flow
Agent processes events into standardized record format
MCP handles protocol translation for each target system
Parallel writes to DynamoDB tables with configurable consistency levels
Optimistic concurrency control with conditional writes
Wallet Creation Process
Wallet creation requests routed through dedicated MCP
Cross-mint service creates blockchain-specific wallets
Response processed and wallet metadata persisted to Wallet Inventory
User-Wallet mappings created with appropriate permission model
Performance Optimizations
Connection Pooling: Configured HTTP and database connection pools
Batch Processing: DynamoDB batch operations for bulk record updates
Read-Through Cache: ElastiCache for frequently accessed wallet data
Write Optimization: Write sharding for high-volume wallet inventory updates
Monitoring & Reliability
Observability: CloudWatch custom metrics with dimension-based filtering
Tracing: AWS X-Ray with annotation for transaction boundary tracing
Logging: Centralized logging with correlation IDs and structured JSON format
Alerting: SNS topics for critical operation failures with escalation paths
Redundancy: Multi-AZ deployment with failover automation
Compliance & Audit
Transaction Logging: Immutable record of all wallet operations for audit
Data Validation: Schema enforcement at API and database layers
Encryption: Data-at-rest encryption with KMS-managed keys
Compliance Hooks: Configurable policy enforcement points for regulatory requirements
Last updated