A virtual machine scheduling algorithm for resource cooperation in a private cloud

Abstract

In recent years, virtualization has been widely applied in cloud computing because of its ability to increase resource utilization. With the scale of cloud computing architecture becoming larger, efficient resource allocation has also become more important. Existing scheduling algorithms for virtual machines cannot use new information to decide upon allocation of the appropriate physical machines because current scheduling algorithms lack the ability to be updated with up-to-the-minute information about each physical machine when making allocations. This situation means a physical machine can be assigned too many virtual machines, thereby causing overloading situations. Therefore, a more efficient and flexible architecture to allocate resources is needed. In this study, we present a cloud architecture and Layered Calculation Virtual Machine Allocation (LCVMA), to perform exceptionally well in terms of achieving above goals. With this architecture and algorithm, we can identify the physical machines with low workloads, and service providers can allow users to use resources more efficiently. The threshold in our mechanism presents possibilities for reducing overload situations. Resource utilization and allocation can therefore become more efficient and economical. © 2012 Springer Science+Business Media.