Design And Implementation Of Intelligent Monitoring Systems For Large-scale Distributed System

The information systems of worldwide large and medium-sized financial institutions have widely adopted distributed technologies to meet their system's parallel high-performance processing and fault tolerance requirements.The enormous number of computing devices comprised in a distributed system with great variety,lead to great difficulties on system operations.In response to the above difficulties,IT operation departments in major financial institutions generally adopt countermeasures,namely,through the construction and implementation of monitoring systems,to achieve automatic failure warning and alarming,performance and capacity threshold identifying,and a comprehensive analysis on the cause of the failure,which ensure IT system operation's safety and stability.Therefore,monitoring systems that meet operation requirements have become an indispensable part of a financial institutions' IT system.An agent is an entity that has knowledge,goals,and capabilities which can operate under certain circumstances.MAS organizes multiple agents organically,through certain communication methods and negotiation mechanisms,in order to complete tasks that a single agent cannot accomplish.Because MAS has good distributivity and flexibility in the design and modeling of system engineering,applying MAS when designing intelligent monitoring systems for large-scale distributed systems can establish an easy-to-expand,high-intelligent,cross-data center monitoring system framework that meets requirements of IT operation for large and medium-sized financial institutions.The design and implementation of intelligent surveillance systems that meets the IT operation's requirements by utilizing excellent features of agent and MAS in design and modeling of system engineering,focusing on the current situation and common problems of monitoring systems in the development of large-scale distributed system,combining with the latest research achievements of information technology,not only has exponential theoretical and practical values,but also has great significance on maintaining stability and improving competiveness for large and medium-sized financial institutions.The main research contents of this thesis are as follows:1.Research in Intelligent Monitoring System Architecture for Large-scale Distributed Systems.A modeling method of the hybrid control for intelligent monitoring system is proposed,based on the monitoring objects,system features,control structures and design principles perspectives of large-scale distributed systems.A hybrid control structure charactered by overall distribution,local concentration and bottom distribution,top concentration is used to divide functional agents and condinate the communication between agents.Thus,the architecture of intelligent monitoring system for large-scale distributed systems is established.2.Functional agent design modeling research.The structural design of functional agent directly influences the overall performance of the monitoring system.The design and modeling research from agent modeling perspective of each acquisition agent,event correlation agent,data bus agent,and display agent are conducted;the design and analyzing of each type of functional agent's structure base on the characteristics of the intellectual monitoring system for large-scale distributed system were represented;and agent structural modeling for each type of agents was completed.3.Research on calculation methods of core monitoring indicators.Obtaining monitoring indicators that closely match the application and business monitoring of large-scale distributed systems is one of the core goals of the monitoring system.From the perspective of the design of system monitoring indicators,the research on the calculation methods of core monitoring indicators was conducted.The core monitoring indicators calculation method was studied based on the specific characteristics of the financial industry applications and services;the calculation methodology for large-scale distributed system availability and robustness focusing on financial industry applications and business characteristics were proposed.4.System implementation.The thesis completed the construction of the intelligent monitoring system for large-scale distributed systems,and verified feasibility and effectiveness of the proposed system model and fundamental theory through engineering practice.The research results of this paper have been fully applied to the operation of IT system of Dalian Commodity Exchange.It has accomplished unified management,authorization,and centralized monitoring for various types of monitoring data and monitoring interfaces across different data centers.