Research On Load Balancing Algorithm For Grain Data Center

The world's population is rapidly expanding and almost every country is very concerned about the food issue.Every country is trying to establish its own food security mechanisms.Without the safe supply of food,our social life will be affected seriously.It will lead to all kinds of panic,and even social unrest.In order to ensure the security of national food supply,we need to build a food data center.We need to accurately control the information of each grain warehouse,such as the type of food purchased,the time of buying food,the time it takes to transport food,warehouse food weight,food quality.At the same time,the collected data can be smoothly transmitted and processed efficiently.As an important part of the national food security system,the food data center is the foundation of the food information platform.With the rapid development of the information industry,the information system is expanding rapidly,and the level of automation is constantly improved.The disadvantage of traditional data management center is obvious,such as low physical host utilization,geographically-restricted business requests,incompletely fault recovery mechanism and so on.The unified national grain network data center based on cloud computing is the development trend in the future.This paper focuses on the key technologies such as virtual machine placement and virtual machine dynamic migration,which are involved in the process of building the national food data center.These technologies provide technical support for building the national food data center.The main research work of this paper includes:1.Analysis of the current situation and the problems that the resources dispatch has to deal with the current situation and the resources deployment of the cloud at home and abroad.The architecture and key technologies of the food data center based on cloud computing are studied.2.Ant colony based on differential hybrid optimization algorithm for virtual machine placement(AOCDEVP)is proposed.The AOCDEVP algorithm is based on an ant colony algorithm.Gaussian kernel function and difference algorithm are introduced in theAOCDEVP algorithm.CPU,memory,and network,three resources are quantified.Using quantized resources,the AOCDEVP algorithmallows for better deployment of the virtual machine,reduces the useless overhead in the data center.3.The dynamic migration strategy based on the predicted virtual machine is proposed.The load balancing situation in the next several cycles is predicted by a prediction function.The adjustment of the virtual machine and the shutdown of the physical host can be performed by predicting functions.The traditional way of adjusting the virtual machine once it's overloaded is avoided,which reduces the overhead of the system.4.Through the Cloud Sim platform,the simulation experiment is carried out.Compared with the greedy placement algorithm(GS)and the ant colony placement algorithm(AOC),the feasibility and validity ofthe AOCDEVP algorithm isproved.