Research For The Policy Of Virtual Machine Dispatcher In Cloud Computing Platform

Virtualization is the foundation of cloud computing platform resource allocation flexibility,Virtual machine resource allocation and scheduling strategy are major bottlenecks affecting the user experience and cloud platform performance. There are resource load imbalance, scheduling delay, difficult issues such as the expansion of traditional randomly assigned polling scheduling mode. I consider load balancing, resource utilization, user experience, optimize the allocation and high reliability, Were studied virtual machine allocation and resource scheduling policy centralized cloud platform and multi-domain distributed cloud platform.I consider cloud computing platform for large-scale centralized resource management difficulties, single point of failure situation, Proposed Producer- Consumer Center-consumer model, the specific resource management decomposed into multiple autonomous systems, reducing management load degree. Centralized cloud platform for large-scale deployment, scalable proposed resource a llocation strategies virtual machine scheduling; Use consistent hashing algorithm, a user requests a resource of Zone(rack) Location; Further use and recycling for the first time polling mechanism to adapt the algorithm to obtain the user requests a virtual machine. In order to overcome resource expansion brings a lot of virtual machine migration issue, and efficient cloud computing platform resource expansion algorithm. Analysis and validation show that the proposed virtual machine scheduling policy can achieve a balanced distribution of resources, improve efficient use of resources.I propose a cross-domain resource management and virtual machine scheduling solution to distributed multi-domain cloud platform deployment, use CDN thinking, consider the quality of the network and cloud computing centers remaining resources, locate the user requests the nearest cloud computing platform, Further centralized cloud computing platform resource allocation scheduling virtual machine is allocated. Analysis and validation show that the proposed allocation model to solve the communication delay and the load distributed cloud computing platform, multi-domain problem of imbalance, distributed cloud platform resources to achieve a balanced, reasonable and efficient management.I focused on the proposed virtual machine resource allocation strategy, design and implementation of the Dragonstack cloud platform management system, this system consists of the resource management module and the virtual machine module. Resource management module is mainly distributed cloud computing platform user requests are directed to the best performing sub-cloud service platform, The specific allocation for each sub-station system management to centralized virtual machine scheduling scheme; Virtual machine using the module is the development of a virtual machine landing software, al owing users to apply virtual machines on demand service.