The Key Technology Research Of Virtual Machine Scheduing Under Openstack

Cloud computing is a new generation business model, which is evolved from gridcomputing, parallel computing and distributed computing. We can use it easily vianetwork and acquire the computing resource on-demand (such as computing resource,storage, network and application service). Cloud computing resource are made upthrough virtualization technology underlying hardware resources virtualization,forming a huge virtual resource pool which is available to the user. so, virtualizationtechnology is the foundation of cloud computing, especially the technology of servervirtualization, is a powerful technology to put IaaS into effect. Currently, the data centeras a basic vehicle to build cloud computing resource pool, approach to the physicalmachine that hosting the virtual machine through virtualization, while the virtualmachine use to host applications. On the cloud computing environment, how to improvethe data center resource utilization and reduce energy cost under the agreement whichmeets the conditions of SLA is an important problem. We want to solve this issue byimproving the virtualized resources allocation and scheduling strategies. The maincontributions and work in this thesis are as follows:(1)We have analysed the basic characteristics of the current cloud computingarchitecture and the key technology. Focusing on the current popular open source cloudcomputing platform OpenStack, we build a distributed OpenStack platform and readthe source code of OpenStack. I’m familiar with the work process and the variouscomponents of OpenStack platform, understanding the working mechanism of thisplatform.(2) Design and implement a system on the OpenStack. It can migrate the virtualmachines dynamically. This system can monitor the resource usage of the virtualmachines in real time and integrate them. It can improve the resource utilization ofphysical servers effectively and reduce its energy consumption.(3) For multi-objective virtual machine placement problem on the cloudcomputing platform, we propose our solution MACS. The algorithm is based on antcolony algorithm, we redesign the model of power consumption and wasteage, so it cando heuristic search more faster. We compare its performance with the algorithm ofMGGA, ACO and FFD in our experiment, we find our algorithm is better on improvingserver resource utilization and reducing server power consumption. At the same time, we test its performance of large-scale data under the Cloudsim simulation platform, italso achieved a good result, so our algorithm is also suitable to big data center.