【作者】 胡博；

【导师】 刘淑芬；

【作者基本信息】 吉林大学 ， 计算机系统结构， 2006， 硕士

【摘要】 随着Internet的应用不断推广,用户对网络的要求也越来越高。为了提供一个良好的网络环境,使网络的运行状态处于最佳,必须有一个高效的网络管理系统对网络进行管理。而数据采集是整个网络管理生产平台进行网络管理的基础和前提,当前复杂的网络结构和各异的数据采集方法对网管平台的数据采集工作提出了挑战。传统的软件开发方法将逻辑过程固化在系统程序代码中,用这种方法开发的数据采集系统无法适应网络的变化。为了适应现代网络的迅速发展,方便网络管理人员管理网络,突破目前数据采集软件发展所处的窘境,我们抛弃了传统数据采集软件开发方法,采用基于规则的数据采集思想,自主研发了基于规则的通用数据采集系统——DCS。本文分析了规则引擎的原理,对基于规则的数据采集系统做了研究和设计,并实现了其中的数据转发子系统。主要包括以下几部分:1、分析了规则管理系统和规则引擎的工作原理、规则引擎的核心算法——Rete算法。对DCS中集成的规则引擎Drools的核心框架做了研究。2、给出了基于规则的通用数据采集系统的设计方案,将数据采集过程中的数据划分、数据处理、数据转发抽象为逻辑规则,将规则引擎集成入数据采集系统中指导数据采集的执行。3、设计并实现了基于规则的通用数据采集系统中数据转发子系统部分,包括子系统中数据转发规则的定制、规则的解析、数据转发方式的具体实现等。更多还原

【Abstract】 With the rapid development of Internet, the network managementsoftware becomes more popular. Because of more and more networkterminals adding into the Internet, the operations of the network becomemore complicated. New equipment is used to improve the efficiency andspeed of the network. Data collection system provides the raw data to thenetwork management system, and it’s the base of network managementsystem. The integrality and accuracy of the data from the data collectionsystem affect the total capability of the network management system. How tocollect, dispose and analyze the data of the Internet exactly and quickly arethe key question. The information change rapidly and increase exponentiallywith the extension of the network, and it’s impossible to collect and disposeall the information. More than that, the structures and the manufacturers ofthe network equipment are various, and their management modes aredifferent, so it becomes more complex for the data collection of network. Inorder to adapt the development of the Internet and break the predicament ofthe data collection software faced, we forsake the traditional developmentmethod and develop a data collection system based on rule engine namedDCS.First of all, we introduced the theories of Rule-Base ManagementSystem, the principles of Rule Engine and Rete arithmetic, the core of RuleEngine in this paper. We analyzed the framework of Drools, a rule engineused in DCS, then brought forward the solution of rule-based collectionsystem, with abstracting the logical rules of data division, data process anddata forward from the process of the data collection. Finally, the data forwardsub-system of DCS is designed and realized, including the customization ofthe data forward rules, the parsing of the rules and the realizations of all theforward modes.The rule-based management technology provides a way to analyze andresolve problems according to the logical rules. The developers pay attentionto the development of the software and the users of the software customizeand modify the rules. All these can make the group worked more effectivelywhen the operations and techniques change. Rule Engine is the core ofRule-based Management System. It tests and matches the data objects to thelogical rules in the rule engine, and then activates the rules which are accordwith the current data state and fires the corresponding operations according tothe execute logic in the matching rules.DCS collects the data on the Internet, analyses and disposes the data,not only provides the data support to TGOP (Telecom Generic ObjectPlatform), but also to other network management products. DCS is made upof these parts: customization of data divide rules, customization of datadispose rules, customization of data forward rules, collection task’sconfiguration, task dispatch, system management, collecting inspect, datacollection and collecting test tools. DCS provided SNMP, COM, Telnet andFTP collection modes. The system have the functions of data structurecustomization, data processing customization and data forward customization,so DCS not only adapts the current network, but also has the ability to extendthe functions of data collection using customizing the new rules. For example,when a new type of equipment added into the network, the current systemcouldn’t recognize the data information of the new type of equipment, butDCS could customize the new data structure division rules of the newequipment to realize the collecting data formatting and semantic explanation.The system also could customize new data processing and data forward rulesaccording to the change of collection requirement, and these functionsensured the changeability requirement of the data collection.The data forward sub-system of DCS is composed of customization ofdata forward rules, rule matching of the rule engine, getting forward data,encapsulation of forward data, data conservation and data forward. In thesub-system, the forward modes are abstract to forward models, and using theexecutable language DRL express the object models through thecustomization of data forward rules, and the these rules are written in the drlexecutable files. All these files make up of the rule base of the system. Therule engine matches the rules in the rule base according to the configurationsof the collection task.There are eleven forward modes which are categorized into four types inthe forward sub-system. They are protocol forward, middleware forward,database forward and file forward. We can customize different forwardmodes to adapt the change of requirement according to different collect task.The protocol forward modes include SNMP protocol’s Trap forward, Telnetforward, FTP forward, TCP forward and UDP forward. The middlewareforward modes include JMS data forward and Java Bean forward. Thedatabase forward modes can be used in Oracle and SQL Server. The fileforward modes include text file and XML file. The data forward sub-systemnot only realized these forward modes, but also provided the extensibleinterfaces to append new forward modes. These interfaces were only used inthe executable files to reduce the system coupling. If rule files’ customizationgenerating interfaces are not provided to users, data forward rule compiledmanually based on the templates can also use the new data forward modedefined in system.The rule-base data forward sub-system was important in DCS. First ofall, it added protocol forward and middleware forward modes rather than theformer data collection system, which intensified the communication betweenDCS and other network management systems. Second, the forwardsub-system provided the interfaces to Telecom Generic Object Platform and itwas the joint of these two systems. Finally, the customization of the dataforward rules was independent on the system, which reduces the coupling ofthe whole network management platform.更多还原