modu-ai Database Domain Specialist
Quick Reference
Enterprise Database Expertise - Comprehensive database patterns and implementations covering PostgreSQL, MongoDB, Redis, Oracle, and advanced data management for scalable modern applications.
Core Capabilities:
- PostgreSQL: Advanced relational patterns, optimization, and scaling
- MongoDB: Document modeling, aggregation, and NoSQL performance tuning
- Redis: In-memory caching, real-time analytics, and distributed systems
- Oracle: Enterprise patterns, PL/SQL, partitioning, and hierarchical queries
- Multi-Database: Hybrid architectures and data integration patterns
- Performance: Query optimization, indexing strategies, and scaling
- Operations: Connection management, migrations, and monitoring
When to Use:
- Designing database schemas and data models
- Implementing caching strategies and performance optimization
- Building scalable data architectures
- Working with multi-database systems
- Optimizing database queries and performance
Implementation Guide
Quick Start Workflow
Database Stack Initialization:
Create a DatabaseManager instance and configure multiple database connections. Set up PostgreSQL with connection string, pool size of 20, and query logging enabled. Configure MongoDB with connection string, database name, and sharding enabled. Configure Redis with connection string, max connections of 50, and clustering enabled. Use the unified interface to query user data with profile and analytics across all database types.
Single Database Operations:
Run PostgreSQL schema migrations using the migration command with the database type and migration file path. Execute MongoDB aggregation pipelines by specifying the collection name and pipeline JSON file. Warm Redis cache by specifying key patterns and TTL values.
Core Components
PostgreSQL Module:
- Advanced schema design and constraints
- Complex query optimization and indexing
- Window functions and CTEs
- Partitioning and materialized views
- Connection pooling and performance tuning
MongoDB Module:
- Document modeling and schema design
- Aggregation pipelines for analytics
- Indexing strategies and performance
- Sharding and scaling patterns
- Data consistency and validation
Redis Module:
- Multi-layer caching strategies
- Real-time analytics and counting
- Distributed locking and coordination
- Pub/sub messaging and streams
- Advanced data structures including HyperLogLog and Geo
Oracle Module:
- Hierarchical and recursive query patterns (CONNECT BY)
- PL/SQL procedures, packages, and batch operations
- Partitioning strategies (range, list, hash, composite)
- Enterprise features and statement caching
- LOB handling and large data processing
Advanced Patterns
Multi-Database Architecture
Polyglot Persistence Pattern:
Create a DataRouter class that initializes connections to PostgreSQL, MongoDB, Redis, and Oracle. Implement get_user_profile method that retrieves structured user data from PostgreSQL or Oracle, flexible profile data from MongoDB, and real-time status from Redis, then merges all data sources. Implement update_user_data method that routes structured data updates to PostgreSQL/Oracle, profile data updates to MongoDB, and real-time data updates to Redis, followed by cache invalidation.
Data Synchronization:
Create a DataSyncManager class that synchronizes user data across databases. Implement sync_user_data method that retrieves user from PostgreSQL, creates a search document for MongoDB, upserts to the MongoDB search collection, creates cache data, and updates Redis cache with TTL.
Performance Optimization
Query Performance Analysis:
For PostgreSQL, execute EXPLAIN ANALYZE BUFFERS on queries and use a QueryAnalyzer to generate optimization suggestions. For MongoDB, create an AggregationOptimizer to analyze and optimize aggregation pipelines. For Redis, retrieve info metrics and use a PerformanceAnalyzer to generate recommendations.
Scaling Strategies:
Configure PostgreSQL read replicas by providing replica connection URLs. Set up MongoDB sharding with shard key and number of shards. Configure Redis clustering by providing node URLs for the cluster.
Works Well With
Complementary Skills:
- moai-domain-backend - API integration and business logic
- moai-foundation-core - Database migration and schema management
- moai-workflow-project - Database project setup and configuration
- moai-platform-supabase - Supabase database integration patterns
- moai-platform-neon - Neon database integration patterns
- moai-platform-firestore - Firestore database integration patterns
Technology Integration:
- ORMs and ODMs including SQLAlchemy, Mongoose, and TypeORM
- Connection pooling with PgBouncer and connection pools
- Migration tools including Alembic, Flyway, and Data Pump
- Monitoring with pg_stat_statements, MongoDB Atlas, and Oracle AWR
- python-oracledb for Oracle connectivity and PL/SQL execution
- Cache invalidation and synchronization
Technology Stack
Relational Database:
- PostgreSQL 14+ as primary database
- MySQL 8.0+ as alternative
- Connection pooling with PgBouncer and SQLAlchemy
NoSQL Database:
- MongoDB 6.0+ as primary document store
- Document modeling and validation
- Aggregation framework
- Sharding and replication
In-Memory Database:
- Redis 7.0+ as primary cache
- Redis Stack for advanced features
- Clustering and high availability
- Advanced data structures
Enterprise Database:
- Oracle 19c+ / 21c+ for enterprise workloads
- python-oracledb (successor to cx_Oracle)
- PL/SQL procedures and packages
- Partitioning and advanced analytics
Supporting Tools:
- Migration tools including Alembic and Flyway
- Monitoring with Prometheus and Grafana
- ORMs and ODMs including SQLAlchemy and Mongoose
- Connection management utilities
Performance Features:
- Query optimization and analysis
- Index management and strategies
- Caching layers and invalidation
- Load balancing and failover
Resources
For working code examples, see examples.md.
For detailed implementation patterns and database-specific optimizations, see the modules directory.
Status: Production Ready Last Updated: 2026-01-11 Maintained by: MoAI-ADK Database Team
Common Rationalizations
| Rationalization | Reality |
|---|---|
| "I do not need an index, the table is small" | Tables grow. The missing index that is invisible at 1K rows becomes a production incident at 1M rows. |
| "I will add the migration later" | Schema changes without migrations are unreproducible. Every change must have a reversible migration script. |
| "This query works fine in development" | Development databases have tiny datasets. Production query plans differ dramatically at scale. Explain analyze first. |
| "NoSQL does not need schema design" | Schemaless does not mean designless. Document structure decisions affect every query and index. |
| "I will just add a column, it is non-breaking" | Adding a NOT NULL column without a default breaks existing inserts. Column additions need default values or migration backfills. |
| "Connection pooling is handled by the framework" | Framework defaults are generic. Pool size, timeout, and idle limits must be tuned to the workload. |
Red Flags
- Schema change committed without a corresponding migration file
- Query uses SELECT * in production code instead of explicit column list
- No index exists for columns used in WHERE, JOIN, or ORDER BY clauses
- Connection string hardcoded in source instead of environment variable
- Transaction scope spans user-facing HTTP request duration (long-held locks)
- No EXPLAIN ANALYZE output for new queries touching large tables
Verification
- Migration file exists for every schema change (show migration file list)
- Indexes exist for frequently queried columns (show index definitions)
- EXPLAIN ANALYZE run for new queries on representative data (show output)
- Connection credentials sourced from environment variables
- Transaction scopes are minimal and do not span I/O waits
- Backup and restore procedure documented and tested
- Connection pool settings configured with explicit size and timeout