We use a detailed simulation study to characterize the performance of bounded voting under a variety of loads and environment, and to compare it to another pessimistic epidemic protocol. We further describe the performance impact of extending the basic protocol to incorporate transparent proxies for planned disconnections.