Significant benefits of Relational database management systems are:
- Powerful operations with data provided by relational data model: first of all is data selection.
- High availability of components, methods and interfaces to access to data from office applications and from programming languages
- Simplicity of databases administration
All benefits just save our time related to development applications, access and operations with data.
In this relation we have three tasks:
- To research if relational model can describe data structures of FSDB
- To research how the operations of FSDB can be described by relational database
- To research how FSDB can be spliced with relational tables
Relation data model
Basic unit of relation data model is a database. Big number of DBMSs allow to operate cross different databases and even within related database servers, however unit of data isolation related to specific database. Also it is important that tables (entities) exist within specific database only.
Databases are created with the use of DDL (Data Definition Language). As we stated before, only DML (Data Manipulation Language) is used for both definition of data structure in FSDB and data manipulation in FSDB. In this perspective we have two alternatives: 1) the use of predefined data structures of RDB for specific FSDB, 2) translate operations of data manipulation to DML: for example, for clustering of FSDB. Option 1) looks like the simplest solution.
With the use of static relational database structure, it should cover structure of FSDB. In the terms of relational database model it represents as one table that refer to itself twice. The first relation organizes tree structure for parent-child branches/signs. The second relation organizes vertical (tree) structure of generator-generated signs.
From point of view of relational model such single-table structure is represented as strongly normalized table. Actually any database with full-fledged structure can be transformed into this strongly normalized structure and vice versa.
Simultaneously strongly normalized data structure can supplement relational data tables (as mixed model of relational and free structured models), as an example, to support storing of sparse properties (attributes).