A Load Balancing Algorithm Based On LPM

Entering the21st century, with the large-scale use of cloud computing, the corepart of the existing network traffic increase, the rapid growth of the traffic and datatraffic, the processing power and computation strength also increases accordingly,makes the single server equipment can not bear such a great amount of calculation.This leads to load balance the concept, i.e., the server load balancing, reasonablyperform share to more than one operating unit, to jointly complete the calculation.Load balancing is a key problem in cloud computing and technology, how toimprove the server load balancing, has long been a research scholars are veryconcerned about a problem. In traditional load balancing strategy, people mainlyadopt the method of real time for the node load information, but the method itself hasvery big defect, real-time information collection node will have delay, make decisionsin a timely manner, the process is not smooth, jitter, to load balance the performanceof the system, the ultimate equilibrium effect is not ideal. Accordingly, if we are ableto accurately or relatively accurately predict node load information, saves the timeconsuming of real-time information collection node load, plus reasonable loadbalancing strategy, will greatly improve the performance of load balance system, so asto improve the performance of the whole cloud computing system.This article first introduces the definition, the principle of load balancing andload collection method, and calculation of the load are analyzed.Secondly analyzes the nature of the load, for below provide a basis for the loadprediction. In this paper, a task scheduling model is established, this model is basedon the cloud computing data centers on the overall architecture. Then give the wholelogic structure diagrams, including input module, load forecast module, selectingmodule, output module, scheduling module that five big modules. We design andimplement the logic chart of every module is given, in the implementation of variousmodules we focus and major classes and functions are given, and explains thefunction of each main class or function, both of each module is given in detail, andinternal diagram of each module is given, the purpose is to better explain how eachmodule.Finally, introduces the cloud computing environment dynamic load balancingalgorithm based on load forecasting, dynamic load balancing algorithm whatindicators, to consider how these metrics available, how to evaluate, and ultimately how to determine quantitative with each task scheduling sequence and execution ofthe host. In view of the task scheduling stage load prediction algorithm is proposedand the node selection algorithm. The virtual machine live migration stage putforward problems, which should be considered and applied to the module architecturemade a brief introduction. At the same time the author studies based on loadforecasting algorithm is introduced in detail, including its main idea, implementation,algorithm process, etc.