Cloud datacenter ($D_c$) have become popular in recent years with the rising popularity and high performance of cloud computing. The multi-step of data computation and diverse task dependencies fail in the task, energy consumption, overloading of Virtual Machines (VMs), and violation of the agreement. To overcome these challenges, we propose a genetic algorithm (GA) based multiphase fault tolerance (MFTGA) approach for intelligently schedule the tasks over the VMs for multiuser. This MFTGA approach efficiently maps optimal VMs with users according to the service level agreement (SLA). The presented approach comprises four phases namely individual phase, local phase, global phase, and fault tolerance phase. In the individual phase of the MFTGA algorithm, we calculate the local fitness $\left(f_l\right)$ of each user. Then calculate the global fitness $\left(f_g\right)$ of multiuser according to the SLA in the global fitness phase. After mapping the optimal VMs with the multiuser, we check the status of task execution in the fault tolerance phase. MFTGA method is used to improve the reliability, latency, and reduce the failure of the task in the cloud computing environment. The proposed MFTGA scheme is compared against the GA and Adoptive Incremental Genetic Algorithm (AIGA). The simulation results validate that the proposed method exhibits better performance than GA and AIGA in terms of execution time, memory utilization, cost, SLA violation, and energy consumption.

云数据中心（$D_C$) 随着云计算的日益普及和高性能，近年来变得流行起来。数据计算的多步和多样化的任务依赖在任务失败、能耗、虚拟机（VM）过载和违反协议方面。为了克服这些挑战，我们提出了一种基于遗传算法 (GA) 的多相容错 (MFTGA) 方法，用于为多用户智能调度 VM 上的任务。这种 MFTGA 方法根据服务级别协议 (SLA) 有效地将最佳 VM 映射到用户。所提出的方法包括四个阶段，即单独阶段、局部阶段、全局阶段和容错阶段。在 MFTGA 算法的个体阶段，我们计算局部适应度$\left(F_{升}\right)$每个用户的。然后计算全局适应度$\left(F_G\right)$根据全局适应度阶段的 SLA 对多用户进行分类。将最优虚拟机与多用户映射后，我们在容错阶段检查任务执行的状态。MFTGA方法用于提高云计算环境中任务的可靠性、延迟和减少任务的失败。将提出的 MFTGA 方案与 GA 和采用增量遗传算法 (AIGA) 进行比较。仿真结果验证了所提出的方法在执行时间、内存利用率、成本、SLA 违规和能源消耗方面比 GA 和 AIGA 表现出更好的性能。