Data generation has increased drastically over the past few years due to the rapid development of Internet-based technologies. This period has been called the big data era. Big data offer an emerging paradigm shift in data exploration and utilization. The MapReduce computational paradigm is a well-known framework and is considered the main enabler for the distributed and scalable processing of a large amount of data. However, despite recent efforts toward improving the performance of MapReduce, scheduling MapReduce jobs across multiple nodes has been considered a multi-objective optimization problem. This problem can become increasingly complex when virtualized clusters in cloud computing are used to execute a large number of tasks. This study aims to optimize MapReduce job scheduling based on the completion time and cost of cloud service models. First, the problem is formulated as a multi-objective model. The model consists of two objective functions, namely, (i) completion time and (ii) cost minimization. Second, a scheduling algorithm using earliest finish time scheduling that considers resource allocation and job scheduling in the cloud is proposed. Lastly, experimental results show that the proposed scheduler exhibits better performance than other well-known schedulers, such as FIFO and Fair. © 2017, Springer Science+Business Media New York.