Research And Implementation Of A Huge Scale Transaction Process Monitoring System

At present, the distributed computing technology has been applied deeply and widely in many key domains. With the scale of the applications in these domains being expanding increasingly, the complexity being increasing and the demand for the reliability being upgrading, the distributed systems take on the following characteristics: the scale becomes huge, the internal relationships are complex, the demand for the reliability is high, the system is hard to manage. So, how to monitor and manage these huge and complex distributed systems to make them have high performance, good availability and good manageability, has become an important task.This article chooses an huge scale transaction processing system as the study background, adopt the distributed object technology based on CORBA, design and implement an monitoring system concerning the huge scale transaction processing system, and introduce its corresponding concepts, principles and key technologies in details. The primary contributions of this thesis include:1. On the base of comparing and analyzing the existed monitoring technologies of the distributed system, this article research and design an architecture concerning the huge scale transaction processing system, and discuss some correlative key problems: the design of Service Agent(SA), the design of Host Manager(HA), data management, policy management, fault analysis and orientation, load analysis and so on. The mode of locating fault base on topology map of object is put forward.2. The monitoring system is implemented based on the StarBus platform. The design and implementation of the monitor and analysis of the statuses of the service objects, application servers, and databases is discussed mainly. We design and implement a host-backup fault monitor, which improves the dependability of monitoring system.The implementation of the load analysis and fault analysis mechanism concerning the huge scale transaction processing system is presented in details. The mechanism of prewarning based on sign reasoning is put forward and implemented.3. The function tests to the monitoring system indicates that the monitoring system implements every function correctly, has little effect to the old system performance when monitoring the old system's inserting status, and enhances the manageability, maintainability and expansibility of the huge scale transaction processing system.