The Cost-Efficient Awareness for Cloud MapReduce

Although the Total Cost of Ownership of the cloud deployment is cheaper than on-premises or colocation, the accumulated costs of usage on the cloud is not an overlooked spending for small and middle-sized business and is particularly vulnerable to spend overhead due to the unmanaged and insouciant increased nodes. Shortly, renting more nodes to improve the performance of cloud computing is not cost-efficient in a pay-as-you-go environment. In this research, we propose the concept of "Cost-Efficient Awareness" to emphasize the importance of cost-efficiencies on the development of cloud economy. To realize this concept, we have derived a cost improvement process within the MapReduce programming model to cooperate with cloud instance provision. We view this cost improvement process as a discrete optimization problem for cloud instance provision under the premise of economic constraint. When we scale up the service, it is critical to keep the costs down. The concept of "Cost-efficient Awareness" just reflects such a consideration. Based on the result of experiment, we can firmly assure the usability of research direction. We hope the future work can make the cloud service more affordable and approachable.