Design And Implementation Of Anomaly Monitoring Subsystem In APM System

With the popularity of distributed technology,the disadvantage of distributed systems that distributed systems are difficult to manage is becoming increasingly apparent,and application performance management is becoming increasingly important to enterprises as a solution for distributed system management.This thesis designs and implements an anomaly monitoring subsystem for realtime anomaly warning based on the APM system involved in practice,and enhances the ability of operation and maintenance staff to sense application anomaly by having them take the initiative to declare the rules for application anomaly.Firstly,this thesis determines the objectives of the system by analysing the operation and maintenance scenarios,and sorts out the system requirements according to the requirement objectives.In addition,this thesis also analyses the problems of anomaly proliferation and anomaly notification storms in distributed systems,and improves the user experience by providing functions such as application anomaly topology and anomaly notification compression.Then,this thesis goes on to design the architecture,functional modules,database and interaction interfaces of the system based on the system requirements.Finally,this thesis implements the system functions according to the system design,and develops a test plan to test the system's function and performance.The anomaly monitoring system implemented in this thesis is a streaming data processing system that efficiently processes application performance data streams by using technologies such as the message queue Kafka,multi-threading,event stream processing engine Esper to achieve real-time alerts.