The Design And Implementation Of Web Service Diagnosis System Based On Business Monitoring Data

Web services are cross-platform and open,and can provide services for all terminals that install browsers.In order to provide services to users more conveniently,some enterprises choose to develop their own web services.Some enterprises lack professional testers.In order to ensure the quality of service,they often outsource the testing work of Web services to testing institutions,which leads to the increase of development cost and the decrease of development speed.How to reduce the defects of Web services under the premise of ensuring the cost and speed of Web services development is one of the important issues concerned by Internet enterprises.In this context,this thesis designs and implements a web service diagnosis system based on business monitoring data,which collect the flow path of user requests in the Web service and running status of each business module through business monitoring.This system can helps developers and operators to diagnose the low-performance methods and interfaces without permission control in Web services.Business monitoring data refer to the log information generated by each business module in the Web service when responding to user requests.The system uses SkyWalking and Spring AOP to collect business monitoring data.SkyWalking collects interface calls in Web services and other data.Spring AOP collects method operation information and other data in Web services.In this system,all system methods are weighted and clustered according to the running time and other parameters,and low performance system methods are obtained according to the clustering results and historical diagnosis results.This system locates the interface without permission control in Web service by diagnosing the anomalies in user access sequences.Firstly,it mines frequent sequence patterns from user normal access sequences,then matches the current user access sequence with frequent sequence patterns,and weights them according to the frequency of each frequent sequence pattern.Finally,according to the matching results,we analyze whether there is an anomaly in the current user's access behavior,and get the information such as the interfaces without access control.The system has been deployed in the test environment of mooctest platform for two weeks.In the test environment,it has been diagnosed that there are seven low performance methods and two interfaces without permission control in the web service.it is proved that this system can help developers and operators to quickly locate the low performance methods and the interfaces without permission control in Web services.