Cloud Resource Allocation Based on Historical Records: An Analysis of Different Resource Estimation Functions

Resource allocation is an important problem for all Cloud Service Providers (CSPs). Some recent studies propose interesting resource assignment models based on the historical behavior of customers. However, they have a few limitations. For example, some of the proposed models are not suitable in all situations or server load conditions. In this paper, we address such limitations from the model in [1] and introduce several new resource estimation functions to achieve better resource allocation. More precisely, four new mathematical models are first proposed and analyzed. Then, we used the CloudSim simulation toolkit to compare the mathematical results and the simulation results. Our preliminary analysis indicates that different models should be used for different situations in order to achieve better resource utilization.