There are several multi-server failure models in the literature. However, no systematic comparisons between models are available. This research tries to compare three types of multi-server failure models by categorizing if there are common failure and repair cycles and if repairman is allowed for vacation. Machine interference problems may arise when no common failure and repair cycles are allowed. We study the aggravated effect of machine interference problems that may arise if repairman is allowed for vacation. We derive and extend necessary mathematic models for our comparative study. Through numeric studies our models provide decision makers quantitative information about the vacation effect. Finally, we state how the optimal policy may be obtained from the viewpoint of operations management.