Data Generation
Configuration
Table row counts, mappings, relationships, and hidden columns.
Data Generation Configuration
Configuration controls both quality and scale of generated data.
Row Count Controls
Global and Per-Table
- Set default row counts across tables.
- Override row counts for specific tables.
Why It Matters
- Too few rows can hide relationship issues.
- Too many rows can slow feedback loops.
Faker Mapping Controls
Each column can be mapped to a faker method.
Typical Examples
- Emails: unique email generators
- Names: person/company generators
- Dates: recent/past/future date generators
- IDs: autoincrement or relation-backed values
Value Profiles
On top of its faker method, each non-key column can carry a value profile (weighted values, patterns, distributions, derived or computed values, coherent addresses). Profiles are created automatically on the first generation and can be changed with the Data Assistant. See Realistic Values.
Relationship Settings
Confirm parent-child pairs when reviewing schema relationships. Foreign keys declared in your schema are always enforced, even if they're missing from the list.
Validation Checklist
- Parent key exists
- Child foreign key exists
- Data type compatibility
Parent Table (Per)
For each child table, choose which parent drives its rows. Combined with row counts this sets ratios: 100 customers and 1,000 orders with orders → customers gives about 10 orders per customer.
Missing or Wrong Relationships
Add application-level foreign keys or remove wrong ones in the Relationships panel or with the Data Assistant. Click Regenerate in the Relationships panel to re-read foreign keys from your SQL. See Relationships and Integrity.
Hidden Columns
Hide columns from exported output when needed.
Save and Reuse
Configuration persists for schema workflows, enabling repeatable runs.
Quality Tips
- Keep enum-like fields constrained; profiles do this automatically with weighted values.
- Use realistic ranges for numeric values; profiles respect column types and
CHECKconstraints. - Select parent tables along with their children, or required foreign keys will stop generation.
- Prefer deterministic patterns for values used in assertions.