In this paper, we propose transmission point selection algorithms to provide lower service time between an application server to end users by selecting the transmission point used by a newly identified metric, i.e., service latency. In particular, the proposed ordered service latency-based transmission point selection algorithm is beneficial in a sense that it does not increase average service time but decrease the maximum service time although it brings additional computational complexity slightly. We also provide the performance bounds with proven approximation ratios and verify that the proposed algorithm achieves lower maximum service time without degrading the average service time as well as waiting time through simulations in practical environments.