Data centers form the infrastructure of information technologies (IT). Most of the Internet Protocol (IP) address, for example, traffic passes through these data centers. Moreover, the IP address data is stored at data centers. To cope with this increasing flow, many processing and storage servers are needed. However, data centers have large investment costs and high energy use. The most important data center cost factors are operating cost and energy cost. Here, there are many factors considered for energy usage in servers and different resource types like CPU, bandwidth, and memory. The CPU, on the other hand, is generally the most energy-consuming part of servers. The most important technology used to save energy in physical servers is virtualization. Virtualization enables many virtual servers (VS) to run on a single physical server (PS). In this study, we aim to meet the resource needs of VSs by minimizing energy consumption in a datacenter. In other words, here we mention how much energy PSs consume according to the amount of CPU used. Here, it is assumed that there is a linear relationship between CPU and energy usage. The workloads of the services running on VSs change in time. With the change of these workloads, the amount of CPU used by virtual VSs’ changes. For this purpose, we propose a stochastic model for server consolidation under resource demand uncertainty. This model, which presents the ambiguous workload, is a scenario-based stochastic programming model. According to the result, in the consolidation of VSs with uncertain resource demand, energy costs are minimized.
Anahtar Kelimeler: Server Consolidation, Stochastic, Modelling, Data center
|
