Information technology is growing fast. The growth of information technology is caused by the increasing of human need for technology. The example of it is cloud computing. The more user who accesses the cloud service can cause the possibility of users being rejected or blocking. The more tasks that are not served so that the performance of cloud data centers becomes less effective. One of method to minimize the blocking probability by controlling the queue system capacity [3]. The system model used in this research is a queuing theory model, load balancing and several physical machines. The queuing model used to analyze is M/G/1/C on load balancing and G/G/m/K on physical machine. General distribution queuing model compatible with the dynamic characteristics of cloud computing. The purpose of this research is to see the effect of queuing system capacity on blocking probability and see the effect of load balancing in this system model. The results of the simulation are the capacity of the queue system which is getting bigger, reducing the possibility of blocking. From this simulation the effective blocking probability value with a queuing system capacity of 3000. The existence of load balancing in this model makes physical machine performance effective. Because load balancing divides the task load equally into each physical machine.