Research On Automatic Deployment Technology For Continuous Software Engineering

In the context of increasing development of technology and increase in the requirements for software use,iterative development,progressive capability acquisition and continuous improvement will become normal for software development.Continue software engineering emerges due to the failure of traditional development methods to meet the needs of current software development.Continuous deployment,a key stage in continue software engineering,plays a significantly positive role in improving the iterative evolution of software.In this paper,a set of automatic deployment frameworks are proposed for shortcomings of software engineering for sustainability such as low deployment efficiency,insufficient deployment fault tolerance and heavy workload of performance parameter configuration.On this basis,automatic deployment strategies and solutions for software performance configuration are studied.This paper mainly focuses on the following content:1)Proposal of an automatic deployment framework for continue software engineering.On the basis of the existing automatic deployment framework,this paper proposes a deployment framework that can address deployment anomalies in an adaptive manner against challenges facing continue software engineering such as high requirements for software iteration update deployment,excessive manual intervention in current software deployment and low reliability of deployment.Specifically,on the basis of studies on the current automatic deployment framework,framework technologies of automatic deployment are proposed together with highly reliable and automatic software deployment procedures,so as to facilitate the improvement of software deployment efficiency.2)Deployment sequence based highly reliable deployment.Regarding to the insufficient fault tolerance of existing software deployment,a software deployment approach based on deployment secquence is proposed,including exception handling method and automatic deployment algorithm.Exception handling method adopts a two-stage exception handling mechanism:In the first stage,before deployment,based on the idempotence and convergence of the deployment sequence,whether the script follows the principles of idempotence and convergence is analyzed to identify script problems in advance;in the second stage,online fault handling methods are provided and a deployment fault pattern library is built systematically to handle deployment faults by means of exception mode processing.Automatic deployment algorithm calculates software deployment strategies automatically based on deployment tasks and software dependencies for the purpose of improve deployment efficiency.3)Recommendation of performance configuration items based on Improved Genetic Algorithm.Regarding to issues such as complicated relationship of configuration parameter items and difficulty in parameter value determination following software deployment,the characteristics and requirements of configuration item parameters of large-scale information system are studied,with a performance configuration item recommendation approach based on improved genetic algorithm proposed.This approach improves genetic algorithm by virtue of real number encoding,adaptive selection strategy and elitism selection mechanism against the deficiency of low efficiency in recommendation of configuration items resulted from complicated dependencies of performance parameters;moreover,it optimizes genetic algorithm fitting with B-SPL serving as a fitness function,so that the accuracy and efficiency of recommendation of performance configuration items are enhanced.