AgentSkillsCN

postgresql-best-practices

全面指导 PostgreSQL 的最佳实践、优化、数据库设计与扩展。适用于当用户寻求数据库设计建议、SQL 优化、索引策略，或 PostgreSQL 性能调优时使用。

SKILL.md
--- frontmatter
name: postgresql-best-practices
description: Comprehensive guide for PostgreSQL best practices, optimization, database design, and scaling. Use this skill when the user asks for database design advice, SQL optimization, indexing strategies, or PostgreSQL performance tuning.

PostgreSQL Best Practices & Optimization

This skill outlines the standards for designing robust, scalable, and high-performance PostgreSQL databases.

1. Database Design (Schema & Normalization)

Naming Conventions (Readability)

  • •Snake Case: Use snake_case for all table names, column names, and indexes (e.g., user_profiles, created_at).
  • •Plural Tables: Use plural nouns for table names (e.g., users, orders).
  • •Singular Columns: Avoid table name prefix in columns unless necessary (e.g., id is better than user_id inside the users table).
  • •Boolean IS/HAS: Prefix booleans with is_ or has_ (e.g., is_active, has_verified_email).

Keys & Constraints (Data Integrity)

  • •Primary Keys: Every table must have a Primary Key (UUIDs are preferred for distributed systems/security, Integers for strictly internal/perf critical tables).
  • •Foreign Keys: Always define FK constraints to ensure referential integrity.
  • •Indexing FKs: FK columns are NOT indexed by default. Index them explicitly if you frequently JOIN on them.

JSONB vs Relational (Optimized Usage)

  • •Prefer Relational: Use columns for structured data that needs independent querying or constraints.
  • •Use JSONB Sparingly: Only use JSONB for truly varying unstructured data or where schema flexibility outweighs strict validation.
  • •GIN Index: If querying inside a JSONB column, create a GIN index on it.

2. Performance & Optimization

Indexing Strategy

  • •B-Tree: Default for equality and range queries (=, <, >).
  • •Unique Index: Enforce uniqueness at the DB level, not just application level.
  • •Partial Indexes: Index only a subset of rows if you often query that subset.
    sql
    CREATE INDEX idx_users_active_email ON users(email) WHERE is_active = true;
  • •Composite Indexes: Use for queries filtering by multiple columns (order matters: equality first, then range).

Query Optimization

  • •Avoid SELECT *: Fetch only required columns to reduce I/O and network overhead.
  • •Avoid OR: OR conditions can sometimes kill index usage. Consider UNION ALL.
  • •Explain Analyze: Use EXPLAIN ANALYZE to inspect query plans for slow queries. Look for "Seq Scan" (Sequential Scan) on large tables—this usually implies a missing index.
  • •Limit & Offset: For large pagination, avoid deep OFFSET. Use specific "seek" (keyset) pagination (WHERE id > last_seen_id LIMIT 10).

3. Scalability

Connection Pooling

  • •Always Use a Pool: PostgreSQL answers processes, not threads. Creating connections is expensive. Use a pooler (like PgBouncer) or the pool built-in to your ORM/Framework.

Read Replicas

  • •Offload heavy SELECT reporting queries to a Read Replica to keep the Primary available for writes (INSERT, UPDATE, DELETE).

4. Maintenance & Manageability

Migrations

  • •Version Control: All schema changes must be versioned migrations (e.g., TypeORM migrations, Prisma migrations, or raw SQL migrations).
  • •Down Migrations: Always write a rollback (down) script for every migration.

Vacuuming

  • •Ensure autovacuum is on to prevent table bloat and wrap-around transaction ID issues.

5. Security

  • •Least Privilege: Application users should strictly have permissions only on tables they need.
  • •No Superuser: Never run the application as postgres superuser.
  • •Encryption: sensitive data (PII) should be encrypted at rest or application level if highly sensitive.