Build a local integration sandbox with synthetic data

An integration project needs repeatable success and failure cases. A sandbox should be small, deterministic and clearly separate from production.

Validate fixture relationships

The integration lab creates synthetic customers, orders and a 503 response.

from integration_cases import sandbox_case

result = sandbox_case()
assert result["referential_integrity"] is True
assert result["failure_available"] is True
assert result["synthetic"] is True
assert result["external_system"] is False

Model the provider contract, IDs, pagination, auth failures, throttling, malformed output and partial state. Seed fixtures and give each test an isolated reset. Provide a fake clock and controllable response sequence.

Synthetic data should preserve relationships and edge shapes without deriving personal values. Label it, review it and keep sandbox credentials separate. Contract tests against an authorized provider sandbox remain necessary because a local fake can drift.

The FDE for Freshers course uses a sandbox before live third-party integration.

Exercise

Build a local HTTP fake with two resources, cursor pagination and scripted 429/503 cases. Add one contract-drift test against a recorded schema.

Continue learning

This article is part of the FDE integration and deployment foundations sequence. Use the neighbouring tasks when you need the prerequisite or the next application.

• Review the prerequisite or neighbouring task in Add idempotency to an order-processing endpoint.
• Continue with Write a Dockerfile for a small Python service.

Reference: FastAPI testing documentation.