Unit testing services in Microservices - Scalability & System Analysis

Scalability Analysis - Unit testing services

Growth Table: Unit Testing Services at Different Scales

Scale	Number of Services	Test Cases per Service	Test Execution Time	Test Infrastructure	Challenges
100 users	5-10	50-100	Seconds to minutes	Local or small CI server	Basic test coverage, manual runs
10K users	20-50	200-500	Minutes	Dedicated CI/CD pipelines, parallel runs	Test flakiness, longer feedback loops
1M users	100-200	1000-2000	10-30 minutes	Distributed test runners, cloud infrastructure	Test data management, environment consistency
100M users	500+	5000+	Hours	Highly scalable test orchestration, containerized environments	Test maintenance, resource cost, parallelism limits

First Bottleneck

As the number of microservices and test cases grow, the first bottleneck is the test execution time. Running all unit tests sequentially or on limited infrastructure causes slow feedback. This delays development and integration.

Additionally, test environment setup becomes complex as services depend on mocks or stubs, which must be maintained and consistent.

Scaling Solutions

• Parallel Test Execution: Run tests concurrently across multiple machines or containers to reduce total time.
• Test Impact Analysis: Run only tests affected by recent code changes to save time.
• Mocking and Stubbing: Use lightweight mocks to isolate services and speed up tests.
• CI/CD Pipeline Optimization: Use scalable cloud infrastructure and container orchestration for test environments.
• Test Result Caching: Cache results of unchanged tests to avoid reruns.
• Incremental Testing: Integrate unit tests with integration and end-to-end tests to balance coverage and speed.

Back-of-Envelope Cost Analysis

• At 1M users scale, assume 150 services with 1500 tests each = 225,000 tests.
• If each test takes 0.1 seconds, total time sequentially = 22,500 seconds (~6.25 hours).
• With 100 parallel runners, test time reduces to ~3.75 minutes.
• Infrastructure: 100 runners with moderate CPU/RAM, plus orchestration overhead.
• Network bandwidth is minimal as tests run locally or in cloud; main cost is compute time.

Interview Tip

When discussing unit testing scalability, start by explaining the growth in services and tests. Identify the bottleneck as test execution time and environment complexity. Then propose concrete solutions like parallelization, test impact analysis, and CI/CD optimization. Finally, mention trade-offs such as cost and maintenance.

Self Check

Your database handles 1000 QPS. Traffic grows 10x. What do you do first?

Note: Although this question is about databases, for unit testing services, the analogous question is: Your test suite takes 1 hour to run. Your codebase grows 10x. What do you do first?

Answer: Implement parallel test execution and test impact analysis to reduce test time and provide faster feedback.

Key Result

Unit testing services scale bottlenecks first appear in test execution time and environment setup complexity. Parallelization and selective testing are key to maintain fast feedback as microservices and tests grow.

Practice

1. What is the main purpose of unit testing in microservices?

easy

A. To deploy the service automatically

B. To test the entire system end-to-end

C. To monitor service performance in production

D. To test small parts of a service independently

Solution

1. Step 1: Understand unit testing scope

   Unit testing focuses on testing small, isolated parts of a service, not the whole system.

2. Step 2: Differentiate from other testing types

   End-to-end tests check the entire system, while unit tests check individual components.

3. Final Answer:

   To test small parts of a service independently -> Option D

4. Quick Check:

   Unit testing = small parts tested independently [OK]

Hint: Unit tests check small parts, not whole system [OK]

Common Mistakes:

• Confusing unit tests with integration or end-to-end tests
• Thinking unit tests deploy or monitor services
• Believing unit tests require full system setup

2. Which of the following is the correct way to mock a database call in a unit test for a microservice?

easy

A. Replace the database call with a mock object returning fixed data

B. Call the real database and check results

C. Skip the database call and do nothing

D. Use the production database credentials in the test

Solution

1. Step 1: Understand mocking purpose

   Mocks replace real dependencies to isolate the unit under test and control test data.

2. Step 2: Identify correct mocking practice

   Replacing the database call with a mock object returning fixed data allows testing without real DB access.

3. Final Answer:

   Replace the database call with a mock object returning fixed data -> Option A

4. Quick Check:

   Mocking = replace real calls with controlled fake ones [OK]

Hint: Mocks replace real calls with fake data in tests [OK]

Common Mistakes:

• Using real database in unit tests
• Skipping important calls without replacement
• Using production credentials in tests

3. Consider this Python unit test snippet for a microservice method that fetches user data:

def test_get_user_data(mocker):
    mock_db = mocker.patch('service.database.get_user')
    mock_db.return_value = {'id': 1, 'name': 'Alice'}
    result = service.get_user_data(1)
    assert result['name'] == 'Alice'

What will this test verify?

medium

A. That get_user_data returns user name 'Alice' using mocked DB

B. That the real database returns user Alice

C. That the database call is skipped entirely

D. That the service raises an error for user 1

Solution

1. Step 1: Analyze mocking effect

   The database call get_user is replaced by a mock returning fixed user data with name 'Alice'.

2. Step 2: Understand test assertion

   The test checks if get_user_data returns a result with name 'Alice', confirming it uses the mocked data.

3. Final Answer:

   That get_user_data returns user name 'Alice' using mocked DB -> Option A

4. Quick Check:

   Mocked DB returns Alice, test checks service uses it [OK]

Hint: Mock return_value sets test data; assert checks service output [OK]

Common Mistakes:

• Assuming real DB is called
• Thinking database call is skipped without replacement
• Expecting error instead of valid data

4. A developer writes this unit test for a microservice method:

def test_process_order():
    result = process_order(123)
    assert result == 'Success'

But the test fails because process_order calls an external payment service. What is the best fix?

medium

A. Rewrite process_order to not call payment service

B. Add a mock for the external payment service call

C. Run the test only when payment service is available

D. Remove the assertion to avoid failure

Solution

1. Step 1: Identify external dependency issue

   process_order calls an external service, causing test failure due to dependency.

2. Step 2: Apply mocking to isolate test

   Mocking the external payment service call isolates the unit test and avoids real external calls.

3. Final Answer:

   Add a mock for the external payment service call -> Option B

4. Quick Check:

   Mock external calls to isolate unit tests [OK]

Hint: Mock external services to avoid test failures [OK]

Common Mistakes:

• Removing assertions instead of fixing dependencies
• Running tests only when external services are up
• Changing production code to fix tests

5. You want to unit test a microservice method that calls two other services: a user service and an inventory service. Which approach best ensures your unit test is reliable and fast?

hard

A. Skip testing this method because it depends on other services

B. Call both real services during the test to check integration

C. Mock both user and inventory service calls with fixed responses

D. Test only the user service call and ignore inventory service

Solution

1. Step 1: Understand unit test isolation

   Unit tests should isolate the method by mocking external service calls to avoid flakiness and slowness.

2. Step 2: Apply mocks to all external dependencies

   Mocking both user and inventory service calls ensures the test is reliable and fast without real network calls.

3. Final Answer:

   Mock both user and inventory service calls with fixed responses -> Option C

4. Quick Check:

   Mock all external calls for reliable, fast unit tests [OK]

Hint: Mock all external services for isolated unit tests [OK]

Common Mistakes:

• Calling real services in unit tests
• Skipping tests due to dependencies
• Partially mocking dependencies leading to flaky tests