| Design patterns are summary of experience which is used to describe the good solution of recurring problems of software development process. Anti-patterns is an extension of design patterns, describing the widely used solutions which have adverse impact on the software system. Performance anti-pattern would have serious implications on the performance of software systems, so research on the effective detection methods of performance anti-pattern are particularly important.In this paper, we aim to detect the performance anti-patterns appear in the software systems, combining the research foundation of laboratory, design and implement a prototype of a software performance anti-pattern detection system. The main contents include the following aspects:(1) Research on the performance anti-pattern. This part of the study includes the collecting and listing of performance anti-pattern from the relevant literature, studied the12kinds typical performance anti-patterns, including describing these12kinds performance anti-patterns using first-order predicate method and collection methods of performance data of these12performance anti-pattern.(2) Proposed the calculation method of performance indicators thresholds based on ARIMA model. In the first-order predicate method of performance anti-pattern, requires the corresponding performance indicators threshold to describe, this paper combining the collected performance data, use the ARIMA model to calculate the threshold of corresponding performance indicators, theory and experiments show that the threshold calculation method is effective.(3) Design and implement the prototype system of software performance anti-pattern detection system. This part includes the overall design of the system, the system is divided into four modules, and four modules are data acquisition module, data processing module, the threshold calculation module and performance anti-pattern detection module, and implement prototype system. |
PDF Full Text Request