The load balancing problem is one of an open issues for the cloud computing. A good load balancing mechanism enhances the performance of network processing, optimizes the use of resources, and ensures that no overloading a single node or link case. The existing load balancing cloud computing research mainly unilateral the fairness of the transmission network or stand only for the system. Hence, this work considers the handling of both network and system load balancing to obtain high performance. To assign the tasks to the same type of nodes along the links with minimum processing and transmission delays subject to the capacities of nodes and links. Three task assignment schemes FCFS, Min-Min, and Min-Max are adopted along with dynamic clustering, which is a method to group the same type of cloud servers. This study changes in the variables manipulated with the number of nodes and the number of tasks and records the maximal end-to-end delay, average end-to-end delay and fairness index, to analyze the load balancing results. The results show that the Min-Max combination with dynamic clustering has a good effect.