The Research Of Highly Reliable SaaS Application Performance Protection Methods Based On Adaptive Software Aaging

With the rise of cloud computing, more and more application systems are released to customers as SaaS model. People relay on SaaS increasingly, and then people require more highly reliability of SaaS. Once the reliability of SaaS applications has problems, the harm caused by these problems is more serious than in the past. The main traditional solutions of the reliability consider network load balancing, software, concurrency, etc. But, application the multi-tenancy and cloud platform architecture of SaaS application make the problems of the reliability even more complex. The most important reason that affects the reliability of Web applications is software aging. Software aging refers to the software system (especially server software with large number of visits and a large amount of data) in continuous operation for some time there will be a performance decline, and even paralysis. In the long process of continuous operation, a number of software errors continue to consume system resources, when resources depletion, the system will inevitably fail, software performance, followed by a recession, and eventually lead to software failure. Practice shows that the complex software system cannot be implemented full coverage test before its introduction. Abnormalities and the programs may lead to misuse can easily be overlooked, which may cause some bad situations in a continuous process,such as memory usage will be leaked, unreleased file locks, the data is not updated in time, storage space debris, accumulation of rounding errors, and specially aging related bugs.The phenomenon of software ageing is very common in SaaS applications. How to deal with the problems of software aging in cluster scheduling to ensure high reliability of Web applications is a hot point in the academic study, and has attracted several well-known international software companies’interests to R&D. But there are no related mature methods. So, the main objective of this study is to shorten the gap between the technical levels of developed countries. To do find the performance protection methods of highly reliable SaaS application based on adaptive software aging, and to resolve the common problems of software aging of web applications. The main contents include:(1) Develop a set of adaptive mechanisms of Saas performance. Based on the traditional system performance protection strategies, analysis the strategic objectives and strategy characteristics several performance protection strategies, Shows calculation of performance indicators of SaaS system overall performance index such as resource utilization and transaction throughput, analysis the basic ideas and processes of the SaaS applications performance protection mechanisms, Form a series strategies including support the system performance evaluation, request admission control and request scheduling of SaaS system adaptive performance protection framework. In order to improve the current of SaaS applications performance protection methods, the paper introduces the theory of fuzzy control and fuzzy evaluation for core requests access and scheduling module. The research are based on request access threshold-based admission control algorithm and server performance evaluation of adaptive level request scheduling algorithm.(2) Put forward a user access intention-oriented research for prediction of SaaS system aging trend. This method regards the user loads as the research object, and indirectly predicts software aging by predicting the users’access intention-oriented. First propose the entire process of this prediction method, and detailed analysis of this process in several key areas. Then focus on how to predict the user’s access intentions, two algorithms are involved here:sequence pattern mining algorithm and pattern matching algorithm, through improving the traditional PrefixSpan sequential pattern mining algorithm and KMP pattern recognition algorithm, to achieve the purpose of predicting user access intent. At last, predict the aging trend of the web server through the visit amount of each page and the page aging damage in a period of time.(3)Propose a grading regeneration strategy of SaaS applications software aging-oriented. Divide SaaS applications into for regeneration levels:operating system level, middleware level, application-level and component-level, and analyzed the advantages and disadvantages of the implementation at all regeneration levels. Then, put forward the framework of software regeneration system, and elaborate the workflow of each functional module. Show the prediction method for software aging, measure the system aging state based on the situation of several resources applications. And determine whether the occurrence of software aging system, the software needed to implement regenerate. After that, this paper proposes regeneration selection algorithm based on multi-attribute decision. Once determining the server needs to implement regeneration, select the appropriate regeneration levels and objects and implement regeneration based on system status and load factors. At last, this paper presents a method for regeneration sequence of components. When implementing component-level regeneration, determine the components of regeneration associated with aging components, according to the coupling between components, All of the components associated with renewable components to a collection of components, components in the collection have the same regeneration order. Then use BP network to determine the set of all components of the regeneration priority, according to the priority order to get a collection of various components of the regeneration of the order, the final order to get the regeneration orders of all components...