As an increasing number of application developers deploy their applications onto the cloud, it is increasingly important to provide a stable and reliable running environment for those application programs, in which it is a challenge that cannot be ignored to do a better job at the scheduling and load balancing of the underlying resource. Under this background, D-Cloud is providing a flexible scaling、resource dynamic scheduling and load balancing as a cloud scheduling platform for D-Ocean, an unstructured data management system.Firstly, this paper summarizes the achievements obtained from the researches of home and domestic to improve the resource utilization of data centers, and proposes a three-tier resource management model of the current cloud computing resource management; Secondly, designs and implements the D-Cloud scheduling platform, which is based on OpenStack. Besides, D-Cloud also references the excellent design idea of the system architecture of it: modularizes all the functions, in order to implement a scheduling platform that centralizes deployment, monitoring, dynamic scheduling and load balancing as a whole, for each module, they communicate with each other through the message queue, which implements a loose-coupled mechanism of service interface with the corresponding implementation, realizes a distributed deploying management, and has an inherent advantage at scalability; Last but not the least, this paper focuses on the improvements and innovations of the cloud computing three-tier resource management model, including virtual machine’s initial placement, dynamic scheduling and physical machine load balancing, in order to make a full use of the computing capacity of the physical resources, to provide a strong support for the resource management of the whole system.Finally, this paper applies D-Ocean system to D-Cloud scheduling platform, in order to verify the ability of D-Cloud to make full use of the dynamic scheduling and load balancing according to the demand of computing resources, which in turn shows the practicability of this cloud scheduling platform. |