Skill: migration-strategies

What I do

• Schema Evolution: Plan and execute schema changes (adding/modifying/removing tables, columns, constraints).
• Data Transformation: Perform data migrations between schemas or systems.
• Zero-Downtime Planning: Implement multi-phase strategies (Expand/Contract) for high-availability systems.
• Rollback Design: Ensure every migration is reversible with tested rollback paths.
• Performance Optimisation: Minimise table locks and use batching for large-scale data changes.

When to use me

• Planning schema changes for production databases.
• Implementing zero-downtime deployment strategies.
• Refactoring database structure whilst maintaining backward compatibility.
• Coordinating schema changes with application deployments.

Core principles

• Safety First: Every migration must be reversible and tested on production-like data.
• Backward Compatibility: Ensure old application versions work during migration phases.
• Incremental Changes: Break large migrations into smaller, safer steps (Expand/Contract pattern).
• Performance Awareness: Use batch processing and non-locking index creation.

Patterns & examples

Batch Processing (Go/GORM)

func (m *Migration) Up(db *gorm.DB) error {
    batchSize := 1000
    for {
        res := db.Exec("UPDATE users SET status = 'active' WHERE status IS NULL LIMIT ?", batchSize)
        if res.Error != nil || res.RowsAffected == 0 { return res.Error }
        time.Sleep(100 * time.Millisecond)
    }
}

Anti-patterns to avoid

❌ Non-Reversible Migrations: Not providing a Down method or rollback path. ❌ Direct Schema Changes: Running AutoMigrate in application startup instead of managed migration files. ❌ Dropping Columns Immediately: Breaking running application versions that still expect the column.

KB Reference

~/vaults/baphled/3. Resources/Knowledge Base/AI Development System/Skills/Delivery/Migration Strategies.md