Research On Method For Network Data Collection And Processing And Formalization

With the rapid development of modern network, the scale and complication of the network are continuously increasing, and the business and service which supports are becoming more various as well, which makes it difficult to manage and maintain the network effectively. And it is a great challenge to manage network effectively and to ensure the running quality of network. One of necessary conditions of well network management is the data of network can be rapidly and exactly collected. Then the data collected can be further processed and analyzed. Consequently, all aspects of network such as the resource, performance and fault can be entirely known. At the same time, the business which network supports and the service quality for customer can also be well known. The data will be provided to network management.In considering on the importance of data collection and processing in network management, research on some relevant problems was conduct. Followings are the main contents of this dissertation:1. According to the research on present network technology and principles of the rule approach, in accordance with some faults of existing data collection and processing methods in network applications, a rule-based method for network data collection and processing was brought forward. And it is from the business perspective on which modern network management focus. In some traditional methods, network data collection and processing mainly were integrated with network management software (the fixed applications). And EJB and Spring are the main ways of structuring the business logic, and it gave us some benefits of configurability and reuse. But the logic code has not the characters of structure and readability because there are plenty of tangled business determination statements. At the same time, it's difficult for the traditional methods to maintain and modify the data processing logic, but the modification is necessary because of the frequent changes of resource and business in network. Especially, the powerful support for important business and customer can not be provided by the traditional methods. So rule-based method was introduced into network management domain, and a rule-based method for network data collection and processing was studied. The logic of network data collection and processing are encapsulated in business rules. First, rule engine was introduced. Any rule system should be established on a rule engine. Rule engine will organize and match rules by some algorithm. Here rules will be organized as a Rete network by an Object-Oriented Rete algorithm. And the rules in Rete network will be matched effectively and rapidly. At the same time, the running rule engine will generate working memory and active queue. Working memory is the main working spot, and the fire of rule and executing of action are completed here. Active queue is used to store the active rules, and some conflict resolution will be adapted to dispatch the execution of rule. Second, rules which encapsulate the business logic were also introduced. And the main parts were illustrated combined with some examples. At the same time, a rule expander mechanism was proposed, and it supports writing rules in natural language-like methods. There are four types of rules: data collection rule, data parsing rule, data processing rule and data transmission rule. Based on the basic principles of the rule approach, the fact model should be built up as the working frame of rule. So the domain terms and facts were defined through the analysis to the characters of network, and the basic fact model which expresses the domain knowledge can be built up. Finally, we illustrated the running mode of rule-based network data collection and processing combined with a whole example. To begin with, four types basic rules are created, and the rules are put into rule engine. Then a rule network is generated, and a working memory is also generated. In addition, after creating collection and processing tasks, the tasks and device entities need to be put into the working memory. Then we can start to collect and process network data. This method is very suitable for current network with great complexity and multiplicity. And the process of data collection and processing is more flexible and effective. The business logic and process are separated by this method. At the same time, the most valuable business rules are structured through an organic manner. Then the rules are able to be read and verified easily. Even network structure and business change frequently, we can also modify the relevant part of rule easily.2. The other point of this thesis is the formal research on network management and data collection and processing, and some problems were rarely referred in previous formal research. A simple formal language was presented, and its semantic was described by CSP. And the language has some characters that are suitable for the formal research and analysis about some problems in network management and data collection and processing. Network data collection and processing is the important part in network management, and it involves a lot of concurrency and communication problems. But some problems have not yet to be adequately addressed are the semantic and behavioral aspects of composition such as how will a composite behave at runtime, will it meet deadline, availability, safety, and security requirements. So the formal research on network management and data collection and processing has been carried on the basis of Hoare's Communication Sequential Processes and theoretical results of network formalization. A simple formal language was proposed, and it has topology style syntax. And this style is very suitable to describe the scene of network. We first introduce topology style syntax including the basic concepts and graphical representation. Then we describe the behavior semantic of the language by CSP including the formal definition of each basic concept and composition semantic between entities. We also extend this semantic definition to address a limited form of compositional minimization. Some of an entity's behavior is irrelevant to composition, because not all interfaces are used. So the unnecessary CSP process descriptions can be eliminated to simplify the model. Some examples were provided, and they all came from the real scenes of network. And we can use the language to analyze these examples. To begin with, the abstraction to the problem is necessary through graphical syntax. Moreover, we need to analyze the semantic of interaction and composition between entities by CSP tool. At last, the formal representation about the whole scene will be generated. Consequently, the method can be used to provide precise formal description and analysis about some problems in network management and data collection and processing. And a reliable mathematical foundation is also provided for the correctness and validity of system design.The creative points of this dissertation are as follows: First, principles of the business rule approach is introduced into the network management domain. In accordance with the characters and requirements of domain, the domain terms and facts were defined. And the basic fact model was built up.Second, this thesis analyzes the creating process of rules of network data collection and processing from forward perspective. And the development process of rules was illustrated combined with some examples.Third, a simple formal language which has a graphical syntax of topology style was proposed. And some of its characters are very suitable for the formal description and analysis to some scenes in network.Forth, some problems in network management and data collection and processing were formally described and analyzed by the language this thesis proposed.In conclusion, in accordance with some existing faults in network applications, this thesis first proposes a rule-based method for network data collection and processing. It is from the business perspective which modern network management focus on, and surrounds the key points of network data collection and processing. It is of great theory significance and practice value. And it also provides effective methods and means for the research on the future network management and data collection and processing. At the same time, we also carry on the formal research on some problems in network management and data collection and processing based on CSP. A simple formal language was proposed, and we can use it to describe and analyze some scenes of network. The right effect also has been acquired. Because of limited time and author's limited capability, the formal research in this thesis just started. But a starting point can be provided for the further work.