A distributed database is formed by a collection of multiple databases logically interrelated in a computer network.
The top-down approach when used in distributed databases correlates a series of stages for the construction of a distributed database project beginning from the ground and is employed in homogeneous systems. The emphasis in the case of distributed databases is given to the data distribution project.This post presents the stages of the top-down approach by means of a a schema, which gives a macro vision of the process. After that the inherent details of each process’s stage are described.
Top-Down approach
The top-down approach is employed in different computer areas. In distributed databases it correlates a series of stages to the construction of a distributed databases project beginning from the ground and is employed in homogeneous systems [1].
The top-down approach is the more usual given the fact that in the majority of the cases there isn’t a distributed database already implemented. If there is a system already established another approach called bottom-up is used.
The picture bellow shows the stages of the top-down approach employed in distributed databases:
As can be seen, in yellow are the traditional stages of the top-down model: requirements analysis, conceptual project, logical e physical project. In red is the distribution project stage that is specific in distributed databases. Note that the physical project in this case is implemented after the distribution project.
The following sections describe the stages of the top-down approach in distributed databases.
Requirements analysis
In this stage takes place the collect of information about the data, and its restrictions and relationships within the organization. The requirements analysis is realized through meetings with the users where is observed how the organization operates. In the end of the analysis a document with the requirements specification is generated.
Conceptual project
In this stage takes place the modeling of the data and its relationships independently of the structure of representation regarding the distributed database system DDS (conceptual modeling). The conceptual project is realized through the analysis of the requirements specification. In the end of the conceptual project a conceptual schema (diagram) with the correct data integrity restrictions is obtained.
Logical project
In this stage takes place the conversion of the conceptual project to the representing schema of a DDS (logical schema). The logical project is realized through the application of conversion rules, translation to the relational model of the distributed database. In the end of the logical project a logical schema with tables, stored procedures, views, access authorizations, etc. is obtained.
Distribution project
In this stage is taken the decision of how the data and programs must be allocated, fragmented through the nodes of the computer network. In some cases the network itself is designed and built to satisfy the necessities of the distributed database project. This stage is considered the most critical in the project of a distributed database.
Physical project
In this stage the logical schema is defined in a DDS suitable to the data model. The physical project is realized by means of SQL instructions. The result is a physical schema in concordance with the established in the distribution project. After finishing the physical project of each node of the computer network the distributed database is ready for use. A monitoring process is initialized and aims to discover possible errors. Such errors are the system feedback and are sent to the people responsible for the construction of the distributed database.
Conclusion
At the beginning of a distributed database project it’s extremely important to assess the organizational environment of the company that holds the data. Obviously when the company doesn’t have a distributed database back-end and legacy systems, the top-down construction approach will be necessarily employed.
As a macro overview it’s possible to infer that in the first stage a document with the requirements is generated and after that the logical and conceptual projects start. Logical and physical schemas start to be generated. With the logical and physical schemas already defined the distribution project starts and this is the most complex. After the definition of the local schema of each node of the computer network, the implementation of the local physical schema starts. At this point each network node is given the responsibility for determined tasks of the company. This means that some objects (views, stored procedures, etc.) of the database are created specifically according to each physical local schema. In the last level is the distributed database monitoring process, which helps in the discovering of bugs and propitiates at the same time the possibility of correction forwarding the bugs to the superior levels.
The top-down approach aims to structure the creation process of a distributed database. Defining and separating the construction stages in a correct manner, the database architects and other people involved in the construction of a distributed database will have more chances of achieving success in a given project. It’ll only happen if the stages are accomplished with strictness and in the established order.
[2] Zhou, Li-Zhu. Distributed Database System Course. 2002. Available at <http://dbgroup.cs.tsinghua.edu.cn/ddb/2002/ddbHandout.zip>. Accessed on November 16, 2007.
[3] Mello, Ronaldo S. Projeto Top-Down de Banco de Dados. 2007. Available at <http://www.inf.ufsc.br/~ronaldo/ine5623/1-introdProjBD.pdf>. Accessed on November 16, 2007.
[4] Ozsu, M. Tamer e Valduriez, Patrick. Notes for "Principles of Distributed Database Systems". 1999. Available at <http://softbase.uwaterloo.ca/~tozsu/ddbook/notes/Design/Design.pdf>. Accessed on November 22, 2007.
Papers
English
http://leniel.googlepages.com/TopDownApproachDistributedDatabases.pdf
Portuguese
http://leniel.googlepages.com/AbordagemTopDownBDDistribuidos.pdf
