Live migrations
Move live data and clients through compatibility, repair, and retirement.
Move live data while preserving authority and history
A backfill copies version 1 after a user has already saved version 2. A deleted record can reappear if the migration treats absence as the only evidence of deletion. Switching readers back is unsafe if new writes exist only in the new store.
Follow the migration method, then drive the local recovery schedules. Name the write authority, versions, deletion records, replay checkpoint, and retirement boundary. Extend the same reasoning to a hot tenant and regional recovery.
Parts group related chapters. Each lesson has a chapter.lesson address, such as 4.07. Open a title below, or use Next to follow the reading sequence. Within a lesson, On this page lists its sections.
- 16.01
Migrate live data while preserving writes, deletions and recovery
Concepts and examples
- 16.02
Reject stale workers and recover uncertain external effects
Concepts and examples
- 16.03
Copy live rows while preserving new writes and deletions
Concepts and examples
- 16.04
Move write authority between shards and account for regional data loss
Concepts and examples
- 16.05
Migrate tenant data with a resumable backfill
Concepts and examples
- 16.06
Design and rehearse regional write failover
Concepts and examples
- 16.07
Erase account data across stores and in-flight work
Concepts and examples
- 16.08
- 16.09
Stage 5: Migrate the reading list to stable tag IDs
Projects and practical assessment