Evaluate related-entity selection and transformed provisioning
Consider it when tests need a coherent customer or order across several systems, with shared provisioning and masking rules. Evaluate the required Fabric and business-entity model alongside the data workflow.
A small disposable fixture may not justify the entity-model and infrastructure setup. The reviewed AI path requires training and documents a single LU/schema, non-hierarchical entity boundary. Do not equate entity snapshots with storage-level virtual databases.
Check evidence and edition scopePostgreSQL dump transformation and subsets, operated by your team
Consider it for a PostgreSQL dump-and-restore workflow where you can maintain transformation rules. Local tooling and inspectable configuration fit teams that already operate their database backups and restores.
Verify sensitive columns, key consistency and uniqueness after restore. MySQL support is beta/in progress. Release 0.2.25 fixes COPY encoding corruption for certain transformed values; upgrading does not repair older affected dumps. Recreate them.
Check evidence and edition scopeJDBC-based relational extraction; assess filters and privacy separately
Consider it when a relational database subset is the deliverable and your team can maintain the extraction rules. The base subset workflow does not require a cloud AI service.
A subset can still contain sensitive data; define and verify filters before export. Check cycles and relationships without declared foreign keys. Version 17.2.5 comparison features do not establish LOB equality in cross-database comparisons.
Check evidence and edition scopeEvaluate virtual copies when refresh and rewind are mandatory
Consider Delphix when point-in-time virtual copies, branching, refresh and rewind are mandatory parts of the test-environment workflow. Assess its separate masking and Synthetic Data products for additional needs.
Check the exact Synthetic Data release, connector support and license entitlements. Perforce announces GA, while one current technical limitations section still says Early Access. That section lists CHECK/UNIQUE enforcement, self/circular foreign keys and cross-schema rule restrictions. Validate those limits for the entitled version rather than carrying forward earlier Oracle-only Preview wording.
Check evidence and edition scopeEvaluate a modular masking, subset and runtime workflow
Best suited to database teams that need reusable masking templates, constrained synthetic rows, deterministic values and separation of project development from deployment.
Privacy should not be credited with Subset product functionality by default; Runtime is separate, runtime deployment is not available for every database, and some workflows require elevated database permissions.
Check evidence and edition scopeEvaluate source-derived masking and relational subsets
Consider it when you need smaller source-derived databases for development or QA, with a CLI-oriented preparation workflow. The documentation names SQL Server, PostgreSQL, MySQL/MariaDB and Oracle; verify the exact engine version.
Subsetting requires foreign-key relationships. Validate application-level relationships that are absent from the database schema. Use a dedicated disposable test environment. A masked copy still needs a separate privacy assessment; successful execution is not an anonymity guarantee.
Check evidence and edition scopeEvaluate code-level values for small developer fixtures
Best for developers who need generated values inside Python or JavaScript tests and can define their own object factories and relationships.
Primarily generates individual values; its documentation says complex objects usually need user-written factories. Seeded output can change across versions, and relative-date methods need fixed reference dates.
Check evidence and edition scopeWhich Tonic product are you comparing?
Structural documentation describes selecting source rows and collecting related records before transformation. Subsetting support varies by connector. Compare Fabricate separately: Reusable Generator workflows and Enterprise private-preview simulations serve generation needs, not the same source-copy workflow. Structural: subsetting · Connector restrictions · Fabricate.
What could make switching the wrong choice?
If the alternative cannot satisfy a mandatory workflow or costs more to operate, switching is a poor outcome even when its feature list looks broader. Include policy migration, connector changes, rewritten scenarios, retraining and rollback in the evaluation. These candidates are editorial starting points, not a complete market inventory or tested ranking.