Improving energy consumption in cloud computing datacenters using a combination of energy-aware resource allocation and scheduling mechanism

Cloud datacenters consume huge amounts of electrical energy resulting in carbon dioxide emissions and high operating costs. In 2013, energy consumed by global datacenters was estimated to be between 1.1% and 1.5% of the worldwide energy usage and is predicted to grow further. This thesis introduces a mechanism for dynamic virtual machines (VMs) consolidation in cloud datacenters. The aim is improving the utilization of computing resources that can decrease the number of activated physical machines (PMs) to decrease energy consumption. The main target is to design a combination of energy-aware resource allocation and scheduling mechanism to decrease the overall energy consumption, and active PMs, besides maximizing resource utilization and minimizing VM migration. In this study, to improve the utilization of cloud resources and reduce the energy consumption of datacenters, a combination of energy-aware resource allocation and scheduling mechanism including DNA based Fuzzy Genetic Algorithm (DFGA) is proposed. By designing a scheduling technique, cloud resources can be allocated efficiently to reduce the energy consumption of the cloud datacenter. Nowadays, DNA plays a vital role in many computing applications due to the massive processing parallelism. In addition, using fuzzy theory in genetic algorithm reduces the iteration of producing the population and assigning the suitable resources to the tasks-based and task length in the node capacity. Therefore, using DNA based fuzzy genetic can obtain the best chromosomes in a few iterations to maximize utilization and minimize VM migration. For subsequent, the energy consumption of cloud computing datacenter is reduced. Energy consumption was analysed in idle and dynamic states of the server, depending on the energy consumed, processes number and size of the data processed, and size of ii the data transmitted for each host.The experimental results were analysed in both synthetic and real Google trace log environments. These experiments were conducted with varying workloads and comparatively analysed through three different metrics: overall energy consumption, resource utilization, and VM migration. The experimental results of applying DFGA algorithm to real Google cloud trace logs show that the energy consumption of the proposed work was 2.15 kWh which was more efficient when compared to other works: Energy-aware Rolling Horizon (EARH) (2.55 kWh), Modified Bit Field Decreasing (MBFD) and Minimization of Migration (MM) (2.65 kWh). The percentage of the system's resource utilization was 82%, compared to other works: EARH (72.8%), MBFD and MM (70%). This study's VMM (X1000) was 2, whereas EARH was 3.2 and MM was 5. It can be concluded that the proposed combination of energy-aware resource allocation and scheduling mechanism can reduce the total energy consumption of the datacenters. The number of activated servers can be minimized by switching off the idle PMs. The resource utilization ratio can be increased and the number of VM migration can be minimized. Future works can apply the proposed mechanism to other cloud platforms.

Preview Text FK 2017 23 - IR.pdf Download (2MB) | Preview

Actions (login required)

View Item