Web Application Front-end CSS Optimization And Back-end Log Analysis

Web application is divided into front-end and back-end. Optimizing thefront-end can improve the performance of Web application and obtain a better userexperience. Analyzing the back-end log file can help developer to detect systemproblems, help them to carry out system maintenance and optimization work. In thepast few years, the automatic analysis and optimization of Web application front-endCSS file and log analysis of the back-end, especially for Hadoop log analysis underthe cloud environment, have gradually become the hot spot of the research field.Based on the studying and analyzing several representative technology shortage ofthe field, this paper proposes corresponding improved method. The specific contentsare as follows:Firstly, the paper introduces some Web application front-end optimizationtechnology, and studies an automatically Cascading Style Sheets (CSS) analysismethod–CILLA which is used to optimize the front-end. During the study, we findthat there will be a high false positive rate when using CILLA to analyze CSS withPseudo-Classes, CSS with attribute selector, CSS declared in the attribute of pageelement, or CSS generated automatically. Through in-depth analysis of Cilla method,we find out the reasons for its high rate of false positives, and propose an improvedmethod. We propose PA（Pseudo class and Attribute）-Match algorithm which is usedto deal with CSS with Pseudo-Classes and CSS with attribute selector, and I（Inline）-Search algorithm which is used to extract the CSS declared in the attribute of pageelement and CSS generated automatically to reduce the analysis false positive rate.The paper gives a comparative experiment, and the experiment result shows theimproved method can reduce20%false positive rate averagely. The new method canbe better on the Web application front-end processing CSS optimization.Secondly, the paper studies the Web application back-end Hadoop log analysistechnology under the cloud environment，and introduces a popular method-SALSA.During the study, we find that there are some shortages in SALSA, such as, it doesn’tanalyze the JobTracker log; it extracts little information from the logs, and thegraphical interface is unfriendly. SALSA uses the idea that is using key Token matchmethod to extract Data-Flow and Control-Flow information from logs to analyzeHadoop logs. We also use this idea to design and implement an improved Hadoop log analysis system–HLAS which has lightweight architecture and friendly graphicalinterfaces. The system can be used to help developers monitoring and analyzing Webapplication back-end environment under cloud environment, guiding them to carryout optimization and troubleshooting. Actual application shows that it can providemore information of the log analysis, guiding the application developer to carry outoptimization, helping them locate back-end fault and troubleshoot easily.