Databases Reference
In-Depth Information
Appendix C
H i s t o r i c a l N o t e s
History is not what you thought.
It is what you can remember.
─W. C. Sellar and R. J. Yeatman:
1066 and All That
This appendix presents a brief and not unbiased survey of some of the seminal research publications in the field of
design theory. The publications in question are listed in chronological order, more or less.
The relational model as such had its origins in two landmark papers by Codd:
E. F. Codd: “Derivability, Redundancy, and Consistency of Relations Stored in Large Data Banks,” IBM
Research Report RJ599 (August 19th, 1969) and elsewhere.
E. F. Codd: “A Relational Model of Data for Large Shared Data Banks,”
CACM 13
, No. 6 (June 1970);
republished in
Milestones of Research─Selected Papers 1958-1982 (CACM 25th Anniversary Issue), CACM
26
, No. 1 (January 1983) and elsewhere.
The first of these papers has nothing to say about design per se. The second, however, includes a section
with the title “Normal Form” that includes the following tantalizing remarks: “Further operations of a normalizing
kind are possible. These are not discussed in this paper.” These remarks appear following an example that shows
how to eliminate relation valued attributes (see the answer to Exercise 12.8 in Appendix D); that's why Codd uses
the phrase “
further
operations” (emphasis added).
Incidentally, the second of the foregoing papers is the source of the term
connection
trap
(see Chapter 9).
Design theory as such began with Codd's introduction of FDs, 2NF, and 3NF in:
E.F. Codd: “Further Normalization of the Data Base Relational Model,” in Randall J. Rustin, ed.,
Data Base
System: Courant Computer Science Symposia Series 6
(Prentice Hall, 1972).
Two brief comments here: First, the title is misleading─further normalization isn't something that's done to
the relational model, it's something that's done to relvars, or rather to relvar designs. (To repeat from the answer to
Exercise 1.1 in Appendix D, the relational model as such doesn't care how the database happens to be designed, just
so long as the design in question doesn't violate any of the precepts of the relational model.) Second, a preliminary
version of some of the material in this paper can be found in two earlier papers of Codd's. The first is an IBM
technical memo, “The Second and Third Normal Forms for the Relational Model,” dated October 6th, 1970. The
second is “Normalized Data Base Structure: A Brief Tutorial,” Proc. 1971 ACM SIGFIDET Workshop on Data
Description, Access, and Control, San Diego, Calif., (November 11th-12th, 1971).
1
Heath's Theorem actually appeared before Codd's 1972 normalization paper, in:
1
This second paper isn't concerned so much with 2NF and 3NF per se as it is with the idea that relations can represent anything that other data
structures─hierarchies, networks, etc.─can. It does discuss 2NF and 3NF very briefly, but its coverage of those topics is essentially limited to
giving a single fairly informal example in each case.
