Integration Testing

Definition

Integration Testing is the second level of testing, where individually tested units/modules are combined and tested together to verify that they interact correctly — checking interfaces, data exchange, and control flow between modules.

Why It's Needed

A module can pass unit testing perfectly on its own, but still fail when combined with others due to:

• Incorrect interface assumptions (wrong data format/type expected)
• Timing/sequencing issues between modules
• Shared resource conflicts (e.g., database access)
• Incorrect data passed between modules

Integration Testing Strategies

1. Big Bang Integration

• Combine all modules at once and test the whole system together
• Simple but hard to isolate the source of a defect when it fails

2. Top-Down Integration

• Start testing from the top-level module, integrating downward
• Lower-level modules not yet integrated are replaced with stubs
• Advantage: major control logic tested early
• Disadvantage: lower-level modules tested late

3. Bottom-Up Integration

• Start testing from the lowest-level modules, integrating upward
• Higher-level modules not yet integrated are simulated with drivers
• Advantage: low-level utility modules tested thoroughly early
• Disadvantage: overall system behavior/control logic tested late

4. Sandwich (Hybrid) Integration

• Combines Top-Down and Bottom-Up simultaneously, meeting in the middle
• Balances the advantages of both approaches

Comparison Table

Strategy	Needs Stubs?	Needs Drivers?	Tests Control Logic Early?
Big Bang	No	No	No (all at once)
Top-Down	Yes	No	Yes
Bottom-Up	No	Yes	No
Sandwich	Yes	Yes	Partially (both directions)

Advantages of Incremental (Top-Down/Bottom-Up) over Big Bang

• Easier to isolate the source of a defect (fewer new modules added at each step)
• Testing can begin earlier, without waiting for the entire system

---

Exam tip: Memorize the 4 strategies with their stub/driver requirements — a very frequently asked "compare integration testing strategies" question.