Data Migration

Data migration changes the source, shape, or location of business data. It needs stronger verification than most infrastructure work because mistakes can become irreversible.

Migration types

Common data migrations include:

• Schema changes.
• Backfills.
• Database engine migration.
• Region or account movement.
• Data warehouse or lakehouse migration.
• Event replay or topic migration.
• Tenant or customer data split.

Planning checklist

Define:

• Source and target owners.
• Data classification and compliance requirements.
• Expected volume and duration.
• Transformation rules.
• Reconciliation queries.
• Downtime or dual-write expectations.
• Rollback or forward-fix path.
• Communication and approval requirements.

Migration flow

Verification

Use more than row counts. Verify:

• Counts by tenant, type, or partition.
• Checksums or aggregates.
• Referential integrity.
• Application reads and writes.
• Permissions and masking.
• Downstream reports and exports.
• Backup and restore after migration.

Watchouts

• Rollback may be impossible after writes diverge.
• Dual writes add complexity and need monitoring.
• Backfills can overwhelm downstream systems.
• Derived data, caches, indexes, and search stores need migration plans.