BPEL服务组合数据流错误的实证研究

发布时间：2018-03-04 04:32

本文选题：BPEL过程　切入点：数据流错误　出处：《南京理工大学》2017年硕士论文　论文类型：学位论文

【摘要】：随着Web服务与云计算的飞速发展,服务组合已经成为应用开发的主流模式之一。Web服务业务过程执行语言(简称BPEL)已经成为描述Web服务组合事实上的的标准。由于BPEL是类XML语言,BPEL服务组合(过程)具有天然的块结构的特性,因而不易发生控制流错误(如死锁)。然而,由于Web服务通过消息进行数据交互,同时为保证数据的隐私性,BPEL服务组合往往需要对内部、外部的数据进行大量的转换处理,因而极易引入数据流错误,这严重影响了 BPEL服务组合的质量。现有BPEL过程的数据流错误检测方法存在路径爆炸和误报(false positives)等问题,为克服上述不足,我们提出了两种方法用于检测BPEL过程中的数据流错误:基于可达定义的数据流分析方法和基于SMT(可满足性模理论)约束求解器的符号编码方法。我们将上述方法应用到178个工业界中的BPEL过程,对其中的数据流错误进行了实证分析,并提出了基于复杂性度量为特征进行数据流错误分类预测。本文的主要贡献包括:1.提出了两种检测BPEL过程中的数据流错误的方法。第一种是基于可达定义的数据流分析方法,第二种是基于SMT求解器的符号编码与约束求解方法。第二类方法首先对常见的三种数据流反模式进行符号编码,然后采用SMT约束求解器对数据流错误进行检测。第一类方法可以方便的定位错误发生的位置,但可能存在误报(false positives),第二类方法可以确保路径的可行性,但需要编码并运行SMT工具,因此两类方法优缺点互补。2.以178个真实BPEL过程为研究对象,对它们的复杂性度量指标和数据流错误进行了实证分析。统计分析了现有度量指标在刻画数据流复杂性方面尚有不足,提出了一组新的度量数据流复杂性的指标。并统计分析发现了数据流错误的严重性,共94个BPEL过程存在数据流错误,占比52.81%,其中输出冗余错误最为严重;最后我们总结出了数据流错误的活动分布和发生原因。3.提出一种以复杂性度量指标作为特征对数据流错误进行分类预测的方法。首先针对常用的复杂性度量指标进行特征筛选,随后基于各种数据挖掘分类算法,对BPEL过程中的各种数据流错误进行分类预测。分类预测均取得了超过90%的准确度且最终选择的特征组合中特征均具有简单、易计算的特点。
[Abstract]:With the rapid development of Web services and cloud computing, Web Services Business process execution language (BPELL) has become the de facto standard for describing Web service composition. Because BPEL is a XML language like BPEL service composition (process). Having the properties of a natural block structure, Therefore, control flow errors (such as deadlocks) are not easy to occur. However, because Web services interact with data through messages, and in order to ensure the privacy of data, Web service composition often requires a large number of internal and external data transformation processing. Therefore, it is easy to introduce data flow errors, which seriously affect the quality of BPEL service composition. The existing data flow error detection methods in BPEL process have some problems, such as path explosion and false positives. in order to overcome these problems, We propose two methods to detect data flow errors in BPEL process: data stream analysis method based on reachability definition and symbolic coding method based on SMT constraint solver. Applied to 178 BPEL processes in industry, The error of data flow is analyzed empirically. The main contributions of this paper include: 1. Two methods to detect data flow errors in BPEL process are proposed. The first method is based on the definition of reachability. The second is symbol coding and constraint solving method based on SMT solver. Then the SMT constraint solver is used to detect the error in the data stream. The first method can easily locate the location of the error, but there may be false positives. the second method can ensure the feasibility of the path. However, it is necessary to code and run SMT tools, so the advantages and disadvantages of the two methods complement each other .2. taking 178 real BPEL processes as research objects, An empirical analysis of their complexity metrics and data flow errors is made, and a statistical analysis is made on the shortcomings of the existing metrics in describing the complexity of data flows. A new set of indexes to measure the complexity of data flow is proposed, and the severity of data flow errors is found by statistical analysis. A total of 94 BPEL processes have data flow errors, accounting for 52.81%, among which the output redundancy errors are the most serious. Finally, we summarize the activity distribution of data flow errors and the causes of data flow errors. 3. A method of classifying and predicting data flow errors based on complexity metrics is proposed. Then, based on various data mining classification algorithms, various data flow errors in the BPEL process are classified and forecasted. The classification prediction achieves accuracy of more than 90% and the features of the final selected feature combinations have the characteristics of simple and easy to calculate.
【学位授予单位】：南京理工大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP393.09

【相似文献】