I had read the Handbook of relational database design by Candice Flemming and Barbara von Halle several years ago. I now return to it after reading several other books on database design.
Flemming and Von Halle (1989, 4) make the following points about the relationship between the logical database model and the enterprise:
Only an organization that understands the information on which the business is built can effectively employ systems technologies to use that information as a strategic asset. Obtaining an understanding of how information is employed within the business and reflecting that understanding within information systems demand a data-driven approach to systems design. In other words, the goal should be to create databases that are not driven by particular access requirements or technological constraints; the databases should first and foremost reflect the information relationships that are important within the business independent of any specific usage.
In this manifesto, the authors are contrasted with Churcher’s (2007) approach of using processes (via Use Cases) to design the database. In this respect, Flemming and Von Halle (1989) are followed by other authors (Mannino (2004) and Oppel (2009)) in positing the extraction of the entities and relationships as the starting point of their database modelling.
At this point of my reading, I have a hypothesis that there are two (2) approaches of how to start database design:
- Process-driven (Churcher 2007)
- Data-driven (Flemming and Von Halle 1989, Mannino 2004, Oppel 2009)
The definition of Logical Data Modelling is given as:
…is a technique for clearly representing business information structures and rules as input to the database design process. (Flemming and Von Halle 1989, 9)
This would appear to be close what Kent (2000, 113) calls the “semantic bridge” between the real world and the database. I will focus on this Logical Data Modelling because that it is what I am pursuing in my study.
The objectives of this modelling (Flemming and Von Halle 1989, 11) are (italics in original):
- “Correct—providing an accurate and faithful representation of the way information is used in the business
- “Consistent—containing no contradictions in the way the information objects are named, defined, related, and documented
- “Sharable—accessible by multiple applications and users to meet varying access requirements
- “Flexible—facilitating additions to reflect new information requirements, “tweaking” to accomodate strengths or weaknesses of a particular implementation approach, and modifications to respond to changes in business operations”
I have constructed the following table to map the objectives for database modelling between the various authors I read so far:
|Flemming and Von Halle (1989, 11)||Hernandez (2003, 28)||Mannino (2004, 127-129)|
|Correct||Accuracy||Ensure Data Quality|
|Consistent||Consistency||Define the meaning of data|
Not all of these mappings are efficient or accurate. This would be an interesting area of research to work out a set of objectives for database design. No wonder the theory of database design is all other the place if there is no agreed upon objectives for the design to meet.
The three (3) schema architecture (Flemming 1989, 18) is similar to that presented in Kent (2000, 28), Middleton (2002, 151), Mannino (2004, 16-17), and Oppel (2009, 7). There seems to be no dispute about the three (3) schema architecture except that it is absent in Churcher (2007) and Hernandez (2003). Its prevalent points to its utility in the database modelling methodologies.
A significant departure for Flemming and Von Halle is that they call each of the user views in the external schema logical data models (LDMs). The relationship to the conceptual schema is:
As the integrated view of all the business’s data, the conceptual schema is effectively a consolidation of all the relevant logical data models. (Flemming and Von Halle 1989, 18)
All in all, Flemming and Von Halle (1989) follow the same methodology as Mannino (2004, 441-455):
A user view is a model or representation of data requirements for one business function. You will eventually combine user views into one integrated logical data model to provide the basis for database design. (Flemming and Von Halle 1989, 111) (Italics in original)
The direction of research on database design would consider the following aspects:
- The objectives of the database design process
- The relevance of the ANSI Three (3) Schema Architecture
- The construction of the external schema through either use-case analysis or data analysis
- The transformation of the external schema into the conceptual schema.
I see these questions as crucial to successful database design. We have to be certain to be know what a good database design looks like. Whether the three (3) schema architecture is applicable has to be addressed. That there are authors who do not address this architecture is interesting. This omission could be the result of ignorance. How to construct the external schema (or its equivalent) is an interesting question as there appears to be several approaches. And how to to transform this external schema into a conceptual model is also of interest.
Churcher, Clare (2007), Beginning Database Design, APress: USA.
Flemming, Candice C. and Von Halle, Barbara (1989), Handbook of relational database design, Addison-Wesley, USA.
Kent, William (2000), Data and Reality, 1stBooks : USA.
Hernandez, Michael James (2003), Database Design for Mere Mortals,
Mannino, Michael V. (2004), Database design, application development, and administration, 2nd Ed., McGraw Hill Companies, USA.
Middleton, Michael R. (2002), Information Management: a consolidation of operations, analysis, and strategy, Centre for Information Studies, Charles Sturt University.
Oppel, Andrew J. (2009), Databases: a beginner’s guide, McGraw-Hill: USA.