Heuristic and task scheduling provide better scheduling solutions for cloud computing by greatly enriching in identifying candidate solutions, ensuring performance optimization and therefore reducing the make span of task scheduling. Several researchers have put forward scheduling and load balancing algorithms for cloud computing systems. However, how to reduce the response latency while efficiently utilizing detection operator mechanisms (switching between groups while scheduling with corresponding task) and reducing communication cost still remains a challenge. In this paper, a hybrid framework called, Multithreaded Locality Task Scheduling and Knapsack Load Balancing (MLTS-KLB) is constructed. The MLTS-KLB first schedules several tasks using Multithreaded Locality Parallel Task Scheduling (MLPTS) algorithm. The MLPTS algorithm gives a definition and method of achieving group synchronization. Secondly, a Knapsack Load balancing model is constructed by extending the migration based model. Then, after formulating the scheduling problems in the MLTS-KLB and bringing forward the MLPTS algorithm based Knapsack Fair Load Balancing algorithm, the efficiency of the MLTS-KLB is validated through simulation experiments. Simulation results demonstrate that the MLTS-KLB framework significantly reduce the latency time of parallel jobs and improves the average throughput of cloud computing environment by minimizing the average task waiting time compared to the state-of-the-art works.