Cloud computing utilizes massive virtualized data centers which consume a large amount of energy. Cloud providers need to reduce the energy consumption of cloud data centers while delivering user requests with a high Quality of Service (QoS) according to the terms in the Service-Level Agreement (SLA). This paper introduces a genetic algorithm for resource allocation (GARA) in cloud data centers that reduces SLA Violations (SLAVs) and manages energy consumption. Simulation experiments based on a real workload show that GARA successfully minimized the SLAVs while maintaining energy consumption at an acceptable level when compared to state-of-the-art resource allocation algorithms.