How do you architect Playwright tests to ensure strict isolation of user session data (cookies, local storage, session storage) across parallel test executions without incurring significant setup/teardown overhead?

Question

How do you architect Playwright tests to ensure strict isolation of user session data (cookies, local storage, session storage) across parallel test executions without incurring significant setup/teardown overhead?

Answer

When designing a robust, high-throughput test suite using Playwright, simply creating a new browser instance per test often introduces unacceptable overhead. However, relying on the default browserContext behavior might lead to unintended session contamination if tests are run in parallel or share resources.

Describe the architectural pattern and implementation details you would use in a large-scale CI/CD environment to guarantee strict, per-test isolation of storage mechanisms (cookies, local storage, session storage) while optimizing for speed. Specifically, discuss:

1. The precise mechanism within Playwright (or the underlying browser driver) that enables isolation.
2. The performance trade-offs between using browser.newContext() versus other potential isolation strategies (e.g., custom fixtures, external state management).
3. How you would implement cleanup logic (teardown) to ensure no residual state leaks between tests, especially when dealing with complex storage like IndexedDB or Service Workers.

Your answer should focus on the engineering decision-making process and the production-grade implementation details, not just basic API usage.