Advanced Database Platform

A Claude Code skill for working with databases at an advanced level. Covers multi-database architectures, query optimization across database engines, data modeling patterns, migration strategies, and operational best practices for managing databases in production environments.

When to Use This Skill

Choose Advanced Database Platform when:

• You need to design a multi-database architecture (SQL + NoSQL + Cache)
• You want advanced query optimization across different database engines
• You need to implement database sharding, replication, or partitioning
• You're planning a database migration (engine change or major version upgrade)
• You want production operational guidance (monitoring, backup, disaster recovery)

Consider alternatives when:

• You need PostgreSQL-specific schema design (use a Postgres schema skill)
• You want Supabase-specific optimization (use a Supabase skill)
• You need a specific NoSQL database (use a database-specific skill)

Quick Start

Copy
# Install the skill
claude install advanced-using-platform

# Design a multi-database architecture
claude "Design the data layer for an e-commerce platform: product catalog, user sessions, order processing, and analytics"

# Optimize cross-database queries
claude "My API queries PostgreSQL for user data and Redis for session data. How do I optimize the combined latency?"

# Plan a database migration
claude "Plan a migration from MongoDB to PostgreSQL for our user service with zero downtime"

Core Concepts

Database Selection Guide

Workload	Best Fit	Why
Transactional (OLTP)	PostgreSQL, MySQL	ACID compliance, relational integrity
Analytical (OLAP)	ClickHouse, BigQuery	Column-oriented, fast aggregations
Caching	Redis, Memcached	In-memory, sub-millisecond access
Document Storage	MongoDB, DynamoDB	Flexible schema, horizontal scaling
Search	Elasticsearch, Meilisearch	Full-text search, faceted filtering
Time Series	TimescaleDB, InfluxDB	Timestamp partitioning, retention policies
Graph	Neo4j, DGraph	Relationship-heavy queries

Multi-Database Patterns

Typical SaaS Architecture:
├── PostgreSQL (primary store)
│   → Users, accounts, subscriptions, core business data
│   → ACID transactions, relational queries
├── Redis (cache + sessions)
│   → Session storage, API rate limiting, hot data cache
│   → Sub-millisecond reads, TTL-based expiry
├── Elasticsearch (search)
│   → Full-text search, filtering, autocomplete
│   → Synced from PostgreSQL via change data capture
└── S3 (blob storage)
    → Files, images, backups, exports
    → Unlimited storage, lifecycle policies

Query Optimization Fundamentals

Technique	Applies To	Impact
Proper Indexing	All SQL databases	10-1000x speedup
Query Planning	PostgreSQL, MySQL	Identify full table scans
Connection Pooling	All databases	Prevent connection exhaustion
Read Replicas	All databases	Distribute read load
Caching Layer	All databases	Eliminate repeated queries
Partitioning	Large tables (100M+ rows)	Faster queries on recent data

Configuration

Parameter	Type	Default	Description
primary_db	string	"postgresql"	Primary database engine
scale	string	"medium"	Scale: small (<1M rows), medium (1-100M), large (100M+)
availability	string	"high"	HA level: standard, high, mission_critical
environment	string	"cloud"	Environment: local, cloud, hybrid
monitoring	boolean	true	Include monitoring setup

Best Practices

1. Start with PostgreSQL, add specialized databases as needed — PostgreSQL handles 90% of workloads well. Only add Redis, Elasticsearch, or other databases when you have a specific performance need that PostgreSQL can't meet. Every additional database is operational overhead.

2. Use connection pooling everywhere — Every database connection consumes memory and resources. Use PgBouncer for PostgreSQL, connection pools in your ORM, and client-side pooling for Redis. Connection exhaustion is the number one production database issue.

3. Monitor query performance continuously — Set up slow query logging with a 100ms threshold. Review the slowest queries weekly. A single unindexed query can bring down a production database under load.

4. Automate backups and test restores — Automated backups that have never been tested are hope, not a backup strategy. Schedule monthly restore tests to verify backup integrity and practice the restore procedure.

5. Use read replicas for read-heavy workloads — Most applications are 90% reads. Route read queries to replicas and writes to the primary. This is the simplest way to scale database performance without sharding.

Common Issues

Database connection errors under load — You're running out of connections. Implement connection pooling, reduce connection timeout values, and ensure connections are released promptly after use. Check for connection leaks in error paths.

Slow queries that were fast before — Table growth changes query plans. As tables grow from thousands to millions of rows, queries that worked fine start doing full table scans. Run EXPLAIN ANALYZE on slow queries and add indexes for the new data volume.

Data inconsistency between databases — When using multiple databases, eventual consistency is expected. For critical operations, write to the primary database first and sync to secondary databases asynchronously. Don't rely on multi-database transactions.