Research On Docker Oriented Micro-services Deployment Strategy

Docker is an open source cloud computing application container engine,and it attracts an extensive attention because it can make a large number of applications running on the existing server.Combining Docker technology with micro services can significantly improve performance,but it also brings about the problem of how to effectively deploy.Based on the multi-objective artificial bee colony algorithm,this paper proposes an algorithm called MOMDA-ABC by using the improved MDA algorithm as the neighborhood search strategy.The algorithm can optimize the communication distance and the host number between Docker containers that deploy micro-services so as to improve the performance of cloud computing platform effectively.MOMDA-ABC(Multi-Objective Marginal Distribution Algorithm ABC),using the MDA to construct the Con group to establish the relationship of variables as a neighborhood search strategy,can solve the problem of the discrete multi-variable correlation multi-objective optimization.And at the same time it also improves the traditional binary coding of virtualization deployment,so that the spatial complexity is reduced from O(N*n*m)to O(N*m),greatly reduces the running time of the algorithm.EDA,based on a combination of group-based evolutionary strategy and probability statistics,can deal with the complex optimization problem well.In this paper the MOMN-EDA algorithm uses the Markov network model as a probability model,and it can make the number of variables in the case of more effective selection of fewer variables for the new individual generation.The experimental results also prove the effectiveness of the method.The significance of this study is that two kinds of multi-objective optimization problems are discussed,which solves the optimization problem caused by the Docker platform,and promotes the deep research of the multi-objective optimization problem as well.It is of a great significance to the practical application of multivariable correlation multi-objective optimization problems.