Research On Virtual Machine Migration In Cloud

With the rapid growth of business application data deployed on cloud data centers by individuals and company deployment,the energy consumption of cloud data centers has become increasingly serious.There are many factors that cause the energy consumption of cloud data centers.Among them,the way of resource allocation determines the task execution mode and the state of physical resource utilization,which directly affects the energy consumption of physical machines,networks,and so on.Inefficient resource allocation ways will cause unbalanced load among physical machines,causing some physical machines to operate at high loads,while others are nearly idling.Using virtual machine migration can cut down energy consumption with effect and realize physical machine load balancing.However,most of the existing virtual machine migration studies overlook the correlation between virtual machines,such as:belonging to the same user,data exchange between virtual machines,etc.The main content of this paper is the virtual machine migration method in cloud environment.The purpose is to adopt the appropriate virtual machine migration method to effectively respond to the dynamic changes of user application load and ensure its performance,further improving the resource utilization of cloud data centers and reducing energy consumption.The main research contents are as follows:1)For network link energy consumption optimization,a traffic-aware virtual machine deposition modus is presented according to the principle that tight communication virtual machines place the same physical machine as possible and reduce network communication.First,construct a weighted communication structure graph between virtual machines.The weight is the communication traffic between virtual machines.The K-means clustering algorithm is used to classify virtual machines.The intra-cluster traffic is tight and the inter-cluster traffic is scarce.The sorting algorithm forms the initial mapping(that is,the initial encoding)of the virtual machine and the physical machine.Finally,according to the fitness function set by genetic algorithm,through a series of operations of selection,crossover and variation,an optimal mapping relationship is finally obtained.Simulation experiments show that the proposed virtual machine migration mechanism achieves better optimization of network link energy consumption and performance.2)A virtual machine migration algorithm based on Federation forming game is presented for load optimization in the scenario of physical machine load imbalance.The algorithm first selects the virtual machine set to be migrated from the high-load physical machine according to the designed virtual machine selection algorithm,and then forms the game theory based on the alliance.The virtual machine set to be migrated forms the alliance structure according to the corresponding constraints.Each coalition in the coalition structure is mapped to a corresponding low-load physical machine.The virtual machine to be migrated continues to perform join and migration operations to update the coalition structure until a stable coalition structure is formed.You will get a better mapping effect,and finally migration based on the final mapping between the virtual machines to be migrated and the low-load physical machines.Simulation result indicates that the migration algorithm realizes better load balancing effect between physical machines.