Design the target schema
- Map source types to target types without losing precision or date range.
- Preserve identifiers and define primary/foreign keys.
- Recreate necessary indexes, uniqueness and check constraints.
- Normalize code pages, strings, NULLs and invalid legacy dates deliberately.
- Separate data cleansing from faithful source conversion.
Load and validate
Stage data before applying relationships; log transformations and rejected records; reconcile row counts, totals and orphan keys; then test the consuming application and transaction behavior. Target SQL syntax alone does not preserve original business rules.
| CSV output | Portable flat delivery without enforced relationships. |
|---|---|
| SQL target | Typed schema, keys, constraints and application compatibility work. |
See CSV conversion, migration services, data recovery and the database pillar.