The Design And Implementation Of Cloud Service Event Anomaly Detection And Analysis System

With the rapid development of cloud computing technology and the increasing demand for cloud resources by enterprises and users,the cluster size and service complexity of cloud service platforms have continued to expand,which has brought challenges to the abnormal diagnosis of cloud service platforms.In the cloud service platform,users execute a series of events to complete the use of cloud service resources.When an abnormal event occurs,viewing the event execution log is the most commonly used anomaly detection method by operation and maintenance personnel.However,in a complex cloud service platform,the execution of an event often involves multiple distributed services,and the execution logs of the events are also distributed in the log files of different nodes and different services,which makes it difficult for operation and maintenance personnel to quickly and accurately use the log to diagnose abnormalities.In order to improve the efficiency of diagnosis of event anomalies in the cloud service platform,a complete set of event anomaly detection and analysis system is needed to help operation and maintenance personnel perform anomaly diagnosis.Aiming at the diagnosis of event anomalies in the cloud service platform,this paper designs and implements a cloud service event anomaly detection and analysis system.According to the requirements of cloud service platform event anomaly detection and analysis,this paper proposes the functions of the system,divides it into log collection and processing subsystem and log analysis subsystem,and separately complete the architecture design,module division and module implementation of these two subsystems.During the diagnosis of event anomalies,the log collection and processing subsystem provides users with flexible log collection configuration items.The user can use this configuration item to customize the initial screening of event logs in the cloud platform.System will extract the unique identifier of the event from the log,and filter the log related to the abnormality of the event according to the identification and the keyword of the abnormal condition.System then clean and store the log.The log analysis subsystem provides analysis of event categories,abnormal link of the event,and abnormal types for the analysis of event anomalies.Among them,the system uses the sliding window method to arrange the collected logs,and enters the logs into a neural network model that combines a One-dimensional convolution network and a Bidirectional GRU network to classify events.Then,the system enters the log into the event process automaton of the corresponding event category to determine the abnormal link of the event.Finally,the system matches the log with the abnormal template knowledge base established in advance,and analyzes the types of event abnormalities.After introducing the design and implementation of the cloud service event anomaly detection and analysis system,this article conducted functional tests and performance tests on the two subsystems to ensure that the system meets the expected design and functional requirements.The test results show that the system has a good effect in cloud service event anomaly detection and analysis.