Cloud computing environments often have to deal with random-arrival computational workloads that vary in resource requirements and demand high Quality of Service (QoS) obligations. It is typical that a Service-Level-Agreement (SLA) is employed to govern the QoS obligations of the cloud computing service provider to the client. A typical challenge service-providers face every day is maintaining a balance between the limited resources available for computing and the high QoS requirements of varying random demands. Any imbalance in managing these conflicting objectives may result in either dissatisfied clients and potentially significant commercial penalties, or an over-resourced cloud computing environment that can be significantly costly to acquire and operate. Thus, scheduling the clients' workloads as they arrive at the environment to ensure their timely execution has been a central issue in cloud computing. Various approaches have been reported in the literature to address this problem: Shortest-Queue, Join-Idle-Queue, Round Robin, MinMin, MaxMin, and Least Connection, to name a few. However, optimization strategies of such approaches fail to capture QoS obligations and their associated commercial penalties. This paper presents an approach for service-level driven load scheduling and balancing in multi-tier environments. Joint scheduling and balancing operations are employed to distribute and schedule jobs among the resources, such that the total waiting time of client jobs is minimized, and thus the potential of a penalty to be incurred by the service provider is mitigated. A penalty model is used to quantify the penalty the service provider incurs as a function of the jobs' total waiting time. A Virtual-Queue abstraction is proposed to facilitate optimal job scheduling at the tier level. This problem is NP-complete, a genetic algorithm is proposed for computing job schedules.