Secretarial knowledge > Graduation design

Research on Distribution Problem Based on Coupling Degree

Abstract: This paper mainly discusses the user distribution, function distribution and data distribution based on coupling degree in distributed application systems. [Keywords] user distribution, function distribution, data distribution, coupling degree, domain knowledge, automatic construction With the complexity and expansion of the Client/Server structure application system, its data, function and user distribution problems have become increasingly prominent. Reasonable and effective design of the MIS system model is the key to the effective performance of the MIS system. Usually these problems are handled by software developers based on their previous development experience. The quality of the design is greatly influenced by human factors, and the solution varies from person to person. This paper proposes a distributed process solution based on coupling degree, and on this basis, implements a distributed automatic generation tool. I. Distribution Problem When we design a MIS system based on the three-tier Client/Server model based on the system's demand information, a very important task is to determine how many clients, application servers, and database servers are served by the MIS system. The composition, how users, functions, and data are distributed across the client, application server, and database server. For a MIS system, we can abstract it into user sets U={u1,u2,..,ul}, function set F={f1,f2,..,fm} and data set T={ T1, t2, .., tn}. The user accesses the data tables in the data set by calling some function modules in the function set. We can illustrate the relationship between them using Figure 1: The distribution problems discussed in this paper include user distribution, functional distribution, and data distribution. User distribution refers to how to allocate users on each platform of the client according to the needs of the system, that is, which users are allowed by each client. Functional distribution, also known as processing distribution, transaction distribution, and logical distribution, refers to how functions are distributed on each platform of the application server, that is, which functions are suitable for which platforms to complete. The granularity of the data distribution can be a database, a data table, a row, a column, etc. The data distribution logic unit we discuss here is a data table, and the data distribution refers to how to allocate data on each platform of the database server, that is, how to Organize this data on the server. Second, the calculation of the coupling degree In the distributed MIS system, there are different clients, application servers and database servers, because there is a similar between users, between functions and between data. Different "attractive". We call this "attractiveness" the degree of coupling. The possibility that we divide two users into the same client is called user coupling. The possibility that two functions are separated into the same application server is called functional coupling. The two data are in the same database. The possibility of a server is called data table coupling. The degree of coupling is discussed here because the degree of coupling is the basis for the division of network nodes in a distributed MIS system. Coupling is a function of different factors. Coupling degree = f user coupling degree The relationship between users, distance and special requirements may affect the user coupling degree. The relationship between the department and the position of the enterprise constitutes a department's job relationship tree or forest, and the leaf node is the user. Two users who belong to the same department directly are more likely to be on the same client than two users who are indirectly in the same department. We define the maximum distance between two users to their nearest common ancestor as the degree of alienation between two users. When two users belong to two relationship trees, their degree of alienation is ¥. In the department job relationship tree shown in FIG. 2, the degree of alienation between users u1 and u4 is 2, and the degree of alienation between users u1 and u3 is 3. The greater the degree of alienation, the smaller the coupling degree. We give different degrees of weight to different degrees of alienation, so that it can be converted into the coupling value. It is more likely that two users are within 400 meters of each other than two users separated by more than one kilometer and the same client. The closer the two users are, the greater the coupling degree, and the users are given different weights according to the distance, which is converted into the coupling value. For some special requirements of the enterprise, we use the degree word to explain the possibility that two users are on the same client. The common degree words are necessary, as much as possible, impossible, impossible, etc. We give different degrees of words differently. The weight of the weight is converted to the coupling value. The degree of functional coupling is obvious. The closer the relationship between all users calling Function 1 and all users calling Function 2, the greater the likelihood that these two functions will be split into the same application server. The relationship of such user-invoking functions has a great influence on the degree of functional coupling, but other factors also affect the degree of functional coupling. The relationship of the data coupling degree function accessing data has a great influence on the degree of data coupling. Other factors affecting the degree of data coupling include the special requirements of the database system and the requirements of the MIS system. Third, the distribution process user, function and data distribution is mainly to establish the Client/Server system structure. We mainly discuss the user, function and data partitioning process based on coupling degree. The user, function and data coupling degree are the basis for our network node division. Users, functions, and data distribution need to solve two problems: 1) How many network nodes are the MIS system. 2) How users, functions, and data are distributed on these network nodes. The solution to these two problems is unified, that is, how to divide the network nodes. Below we take the user distribution as an example to illustrate the specific steps of the distribution process. Set the user set U={u1, u2,...um} of the enterprise, and Lu is the lower limit of the user coupling degree of the client grouping. Our goal is to find a minimum number of client sets C={C1, C2, C3,..., Ck}, client C1, C2, C3,..., Ckí U, satisfy C1, +C2+ C3+...+Ck = U. Any Ci C, Cj ∈ C, i ≠ j has Ci ∩ Cj = φ. If RF>=Lu, the user ui, uj are on the same client. The specific steps are as follows: The first step is to calculate the degree of coupling between users. Here, RFd, RFc, and RFs represent the degree of coupling between users ui and uj due to factors such as distance, membership, and special requirements. RF = RFd + RFc + RFs +... The second step is to find the user binary relationship Ru. Ru={<ui, uj > | " ui, uj ? U and RF3 Lu } The third step is to find the equivalence relationship of Ru Ru?Ru? t the fourth step, the client divides C. C=U/ Ru? /P>Distribution is a trade-off process. There is no single correct solution. It is necessary to balance the objectives and constraints, and comprehensively consider all aspects. Fourth, the automatic generation tool of distribution We designed and implemented a MIS system auxiliary development environment NDTool, This is an automatic generation tool for user, function and data distribution based on coupling degree and domain knowledge. The system design process runs through our ideas and methods. ECSL is an enterprise concept structure description language, mainly used to obtain end users. Information on the needs of the company provided. It is in the natural language-like form of Chinese and is intended for end users. DKDL is a domain knowledge description language, which is mainly used to acquire the experience accumulated by domain experts in the development of MIS systems, and the fields obtained. Knowledge is stored in the domain knowledge base. The language is oriented to domain software experts. The domain knowledge base is used to store domain knowledge. The development experience of domain experts acquired by DKDL is divided into , organization, the results obtained are stored in the domain knowledge base. We provide a set of domain knowledge base operation interface for the access and maintenance of the domain knowledge base. MADL is a distributed MIS system architecture description language, mainly It is used to describe the distribution of users, data and functions in the MIS system. It is the final result of the NDTool system. The language is for software developers. The workflow of the NDTool system. First, the domain knowledge mastered by domain software experts, Described as DKDL text, compiled by DKDL compiler, and called the domain knowledge base interface to store the analyzed results into the domain knowledge base. When the MIS system needs to be developed, the enterprise end user describes his demand information in ECSL language. Come out. Compile with ECSL compiler, refer to the domain knowledge of domain knowledge base, calculate user, function and data coupling degree, then divide network nodes according to coupling degree, and finally describe the result in MADL language. Provide for development Personnel conduct data model and module design. V. Conclusion This article revolves around coupling The distribution process of the Client/Server structure application and the implementation of its automatic generation tool are discussed. The distribution problem has an uncertain feature, and the distribution will consume a lot of energy from the developer. How to design the distribution construction method and design reasonably The distributed automatic generation tool is an important issue in the development process of Client/Server system. References: [1] Lu Yi, Jin Zhi, Wan Ronglin, Xia Youming, Journal of Software Science, Vol.7, No. 3,1996[2] Zhao Hongjun, Zhou Lizhu, Processing Distribution Computer Science in Client/Server Architecture System, Vol. 24, No.5, 1997[3] Zhang Zuoliang C/S Application System Design New Concept and New Technology Computer Science , Vol.24, No.1, 1997

recommended article

popular articles