The stationary number of customers in the infinite-server system with non-Poisson arrivals is dependent on the form of the service time distribution. Using the second law of thermodynamics, we prove the following paradoxical behavior: when the inter-arrival time is a Decreasing Failure Rate (DFR) random variable, the stationary number of customers in the system becomes stochastically less variable, as the service time becomes stochastically more variable. Next, we also study the loss system. When the inter-arrival time is a DFR random variable, the more variable service decreases the blocking probability. In particular, when the number of servers is equal to one, we prove the fact without the help of the second law of thermodynamics.