Synthetic Test Data Vocabulary Exercises

1 / 5

A QA engineer says: "We use test data management (TDM) to ensure our test environments always have compliant, representative data."

What does Test Data Management (TDM) encompass?

Test Data Management (TDM) — the discipline:

TDM activity	Description
Data sourcing	Select which real data (or generate synthetic) for test env
Data masking	Obfuscate sensitive fields in copies of real data
Synthetic generation	Create artificial datasets that avoid real PII
Subsetting	Extract a representative slice of production data
Compliance governance	Ensure test data meets GDPR, HIPAA requirements

Key vocabulary: test data provisioning, test environment parity, data refresh cycle.

2 / 5

A team debates using synthetic test data vs. masked real data for their staging environment.

What is the key difference between the two approaches?

Synthetic vs. masked test data comparison:

Attribute	Masked real data	Synthetic test data
Origin	Derived from real records	Generated from scratch
Re-identification risk	Residual risk remains	Low (if well-designed)
Realism	High — reflects real patterns	Depends on generation quality
GDPR compliance	Requires careful masking audit	Easier to argue non-personal

Key vocabulary: data masking, pseudonymisation, anonymisation, residual risk, de-identification audit.

3 / 5

A data engineer says: "Our synthetic test data generator maintains referential integrity across the order, order_item, and customer tables."

What does maintaining referential integrity mean in synthetic test data?

Referential integrity in synthetic test data:

Constraint	What the generator must ensure
Foreign key (FK)	FK value in child table exists in parent table
Cardinality	One customer can have many orders (1:N)
Ordering	Parent record generated before child record

Tools: Faker with relationship tracking, Gretel.ai, Tonic.ai, DBUnit. Key vocabulary: parent-child relationship, dangling foreign key, generation order, topological sort of tables.

4 / 5

A DevOps engineer implements subsetting production data for the staging environment.

What is production data subsetting in the context of test data management?

Production data subsetting — what it achieves:

Goal	Technique
Smaller dataset	Take 5% of customers, pull all their related records
Referential integrity	Cascade-select related tables
Coverage	Include edge cases: free-tier, enterprise, churned accounts
Compliance	Mask PII before copying to staging

Key vocabulary: representative subset, edge case coverage, cascade selection, staging environment parity, data refresh automation.

5 / 5

A compliance officer asks: "Is our staging environment GDPR-compliant? Are we using real personal data in test?"

What makes a test environment GDPR-compliant regarding test data?

GDPR-compliant test environments — requirements:

Approach	GDPR status
Real production data, unmasked	Non-compliant — no legal basis for test processing
Real data + masking	Acceptable if re-identification risk is demonstrably low
Synthetic data (no real records)	Cleanest approach — not personal data if not derived from real records

UK ICO and European EDPB guidance: using real personal data in test systems requires a lawful basis and strict controls. Key vocabulary: data minimisation principle, purpose limitation, Article 5 GDPR, privacy by design, DPA 2018 (UK).