The limited computing ability of mobile device constrains its performance on complex mobile applications. Mobile cloud computing (MCC) has therefore emerged to migrate computation-intensive tasks to remote clouds or mobile cloudlets. Most strategies allocate tasks with minimal response time yet few consider the load of the nodes. In this paper, we focus on the load balancing problem for nodes when they conduct offloading. We first establish a five-tuple characterized task model to capture the response time of offloaded tasks. Then, we formulate the task allocation problem as an integer linear problem (ILP) under certain conditions. Furthermore, we propose a two-step appointment- driven strategy to solve this problem with minimal task response time. Specifically, a modified genetic algorithm (GA) is adopted to coordinate the load of the nodes. Simulations are conducted to prove the feasibility of our strategy and evaluate the performance of load coordination.