Test Data Mgmt

Challenges

• Real prod data có PII → privacy violation.
• Synthetic data lack edge case real data has.
• Test depend on specific data state.
• Data refresh blow up parallel test.

Strategies

1. Test data factory

const userFactory = {
  build: (overrides = {}) => ({
    name: 'Khang ' + Date.now(),
    email: khang${Date.now()}@test.vn,
    role: 'admin',
    ...overrides
  })
};
const u = userFactory.build({ role: 'guest' });

2. Database fixtures

• Seed file run before suite.
• Use transaction rollback to reset.

3. API setup endpoints

• POST /test/users — create test user.
• DELETE /test/users/:id — cleanup.
• Only enabled in non-prod env.

4. Snapshot / refresh

• Weekly anonymize prod → staging.
• Tools: Delphix, Tonic, Mockaroo.

VN compliance

Luật ATTT, Personal Data Protection Decree 2023:

• Mask name, phone, ID number.
• Salt hash password.
• Strip address detail (keep city only).

Anonymization techniques

• Pseudo — replace name with "User1, User2".
• Generalization — exact age → age range.
• Perturbation — small random offset to salary.
• Synthesis — generate fresh fake data.

Anti-pattern

• Share staging account in Slack → security risk.
• Hardcode test account ID in test code → fragile.
• Cleanup miss → DB grow infinite over months.

Idempotency

Test should leave env in same state as start. Even when fail (use try/finally cleanup).