The Research Of Load Balancing Technology In Cloud Computing

Load balancing is one of the most important technologies in the cloud computing model.A large number of servers could work together to show the powerful calculation capabilityneeds the support of load balancing technology. Load balancing technology play a big role inbalancing load distribution of the cloud computing system, improving the computingefficiency, reducing the task response time, improving the system robustness and so on. Thewide application of cloud computing makes load balance technology become a hot researchamong scholars. Although the use of intelligent optimization algorithm makes the cloudcomputing load balancing has a lot of research results, these achievements mainly used forpast cluster system load balancing technology. According to the characteristics of cloudcomputing to research load balancing technology is still in the exploratory stage. This paperdeeply research the load balancing technology of cloud computing. According to thecharacteristics of cluster and virtualization, by the task scheduling and virtual machinemigration two levels to achieve load balancing of cloud computing system.When hybrid dynamic load balancing algorithm is applied to cloud computing, someproblems occurred, such as sites information exchange frequently lead to low processingefficiency and algorithm is lack of fault-tolerant mechanism. Hence this research proposes aload balancing strategy based on multi-layer and fault-tolerant mechanism. Improvedmethods to solve the problem of initial site did not consider the load, it leads to the loaddistribution unbalance and reduce the load difference between the sites in the process ofinitialization. When choosing site scheduler according to the load on the server, avoid notconsidering load lead to choose the server whose lead is too large or too small, influence theoperational efficiency of the site. This algorithm fuses the advantages of centralized anddistributed strategy. By organizing neighbor sites, the sites information exchange is controlledwithin a certain range. When sites schedule tasks, it appends the load information of itself andits neighbors to the job transfer request. At the same time it is introduced into fault tolerancemechanism without increase system load. It makes full use of idle resources in cloudcomputing system to backup tasks and improves the system reliability.According to the analysis of load balancing strategy of resource scheduling layer, thePolicy-based Agents for virtual machine Migration strategy exists the problems of deployingtoo much agents and using single threshold to trigger virtual machine. By using systemhierarchical idea, this research proposes virtual machine migration strategy based onhierarchical management. This strategy divides the cloud computing system into a number ofsites and deploys a migration manager for each site to load management and virtual machinemigration. It reduced the resource spending of the multi-agents method. This strategyimproves the mechanism of virtual machine migration triggering and brings in forecast ofload incremental extremum. It triggers virtual machine migration by forecasting the extremumof load. It can reduce unnecessary migration and resource spending. This strategy designedthe detailed message response mechanism to prevent the loss of information in the migrationprocess. By using the method of iteration to migrate the virtual machine memory, it reduced downtime in the process of virtual machine migration.