| In recent years, cloud computing has quickly become a widely acceptedcomputing model. As the core concepts of cloud computing, the industry andacademia have a certain consensus, and new business mode and consumptionpatterns that cloud computing provided have recognized.With the characteristics of dynamic scalability, easy to use simple,on-demand service. Cloud platform as an ultra-large-scale cluster, load balancingsystem is a core technology of cloud computing for ensuring cloud platform to bestable, reliable, and take full advantage of platform resources. Currently, lack of asuitable load acquis ition framework for cloud platform, and in such large-scaleresource node environment, how to balance legitimately based on load informationsystem, and the balance strategy also needs to reconsider.In this thesis, through analys is and research on the above series of problems,we will present a resource-based load balancing system. Through abstract cloudcomputing services and specific to the form of a cloud platform, we put forwardthe five level resource model, followed by the phys ical server, a cluster of virtualmachines, virtual machine, virtua l sessions and virtual applications. Taking intoaccount the heterogeneity of the resources in the cloud platform, we make thevarious resources isomorphic by the theory of rough sets, thereby it will reduce thecomplexity of the manage ment of resources.The resource monitoring and load regulation of load balancing system isbased on the five leve l resource model, and use a evaluation strategy that takes theinherent resources of the machine load balancing information capacity and systemresources remaining into load factors. System through the combination taskscheduling and resource migration balances the load of cloud platform system.Service request scheduling based on current load information can achieve theequilibrium of the load to a certain extent as much as possible, thereby it reducesthe work of resources migration and load balanc ing. |
PDF Full Text Request