Discovery: Data Migration from Shared to Isolated

[2witstudios]

Overview

Plan the migration from current shared database to per-tenant isolated databases.

Key Questions

1. How to extract existing team data from shared DB?
2. Schema management: Same schema per tenant or customizable?
3. Migration tooling: pg_dump/restore, custom scripts?
4. Cutover strategy: Big bang or gradual?
5. How to handle cross-tenant features (if any)?

Current Data Model

• drives (shared, with ownerId)
• pages (shared, with driveId FK)
• chat_messages (shared)
• users (shared across teams)

Deliverable

Migration playbook with rollback plan.

Current State

The control plane (apps/control-plane/) handles tenant provisioning with per-tenant Docker projects and encrypted secrets. Multi-tenant isolation tests exist at packages/lib/src/__tests__/multi-tenant-isolation.test.ts. This discovery is about the actual data migration path from shared to isolated databases.

Related

• Depends on: Multi-tenant Infrastructure Isolation (Discovery: Multi-tenant Infrastructure Isolation Architecture #772)
• Blocks: AI Sandbox implementation (Discovery: AI Sandbox with Full Database Access #773)