Design And Implementation Of Distributed Services Monitoring System

As SOA(Service Oriented Architecture)gradually matured,many large companies began deploying SOA internally to improve reusability and maintainability of various information systems.Therefore,runtime monitoring of distributed services in SOA is a meaningful problem.This paper presents a monitoring system for monitoring the QoS(Quality of Service)and system performance of distributed services at runtime.This monitoring system is able to calculate and collect both operating system level metrics and application level metrics,then implement metrics data collection and storage of complex distributed topology.For the latest metrics data collected,this monitoring system provides a visual graphical user interface and automatic alarm function for metric overrun.First,the background theory and related technologies are introduced.Then,according to the characteristics of distributed services,a detailed needs analysis is finished from both functional and non-functional aspects.Next,on the basis of the needs analysis,the outline of monitoring system is designed,the framework of the system is presented in picture,the basic responsibility of each module is introduced.After that,specific implementation details for each sub-module of the most important QoS metrics module are described.Then,the method to implement metrics data collection,storage and visualization of multi-level distributed complex topology by using Ganglia is introduced.Furthermore,the method to implement metric overrun check and e-mail alarm function by combining Ganglia and Nagios is introduced.Finally,a test system is set up and a series of test cases are designed to test each functional module.Analysis for visual metric graphics is proposed to prove the actual meaning and value of this monitoring system for distributed services.At the end of this paper,design and implementation of distributed services monitoring system are summarized,the direction of future work is determined.