开源SDN控制器并发缺陷的量化分析研究

Quantitative Analysis and Research on Concurrency Bugs of Open Source SDN Controllers

SDN(software-defined networking)控制器是SDN网络的核心软件,它的并发缺陷会使网络的运行出现不确定甚至错误.本文对开源SDN控制器的真实历史并发缺陷报告进行深入实证量化分析:首先,从缺陷管理工具自动获取真实缺陷报告并进行预处理获得数据集;其次,基于软件缺陷生命周期选取4个维度10个具体度量作为并发缺陷修复过程的评估指标,基于指标对数据集进行修复过程的特征提取,自动获得其特征集;再次,基于特征集给出包含5种并发缺陷类型的数据集;最后,采用量化统计分析方法对于具有修复过程特征的并发缺陷数据集进行分析,给出SDN控制器在并发缺陷类型、功能模块和修复过程的数据分布特征及分析结果,并进行了有效性讨论.

SDN(software-defined networking)controller is the core software of SDN,and its concurrent bugs will make the network operation uncertain or even wrong.The real history concurrency bug reports of open source SDN controllers are empirically and quantitatively analyzed in this paper.Firstly,real bug reports are automatically obtained from bug and issue tracker and preprocessed to obtain raw data set.Secondly,based on the software bug life cycle,four dimensions and 10 specific metrics are selected as the evaluation metrics of concurrent bug repair process.Then,repair process features of the raw data set are extracted based on the metrics,and a feature data set is automatically obtained.Thirdly,a data set containing five concurrent bug types is given based on the feature data set.Finally,quantitative statistical analysis method is used to analyze the concurrent bug data set with characteristics of repair process.Data distribution characteristics and analysis results of SDN controllers in concurrent bug types,functional modules and repair process are given,and effectiveness is discussed.