COMPARISON AND MAPPING OF THE RELATIONAL AND CODASYL DATA MODELS -- AN ANNOTATED BIBLIOGRAPHY Gary H. Sockut Institute for Computer Sciences and Technology National Bureau of Standards Washington, DC 20234 Two of the best-known data models are the Computing Surveys, ACM, vol. 8, no. relational model and the CODASYL model ], Mar. 1976, pp. 43-66. (sometimes called the DBTQ, network, or data structure set mode]~ . There have E. F. Codd, "A Relational Model of been many efforts at comparing the two Data For Large Shared Data Banks", mode3s and at defining mappings between Commun. ACM, vol. 13, no. 6, June them. The publications 3isted in this i~, pp. ~77-387. bibliography describe some o~ these efforts and some related topics. I invite For background on the CODASYL model, see readers' suggestions for additional sources such as: entries for possible inclusion in a future bibliography. % omit publications on the CODASYL Data Description Language more general topic of data base Committee, "Journal of Development", conversion. Materiel Data Management Branch, Dept. of Supply and Services, Following each reference, I describe the Canadian Govt., Hull, Que., Canada, publication briefly, usually by quoting Jan. 1978. parts of the author's abstract and/or main text. Often I supplement or replace the C0DASYL Programming Language author's text with my own words, enclosed Committee, "Data Base Task Group in brackets ([]). I do not guarantee that Report", ACM, Apr. 1971. an author's abstract accurately reflects the content of the publication. R. w. Taylor and R. L. Frank, "CODASYL Data-Base Management I do not evaluate the publications, nor do Systems", Computing Surveys, ACM, I guarantee the truth of their statements. vo]. 8, no. I, Mar. 197~, pp. 67-103. Each publication may reflect its author's opinions. Also, because specifications For background on both models, see sources have evolved over the last ten years, some such as: statements may be out of date. The bibliography reveals that the authors have C. J. Date, An Introduction to not reached consensus on equivalence of Database Sys'-~ems, 2nd ed~, the models, nor on the models' relative Addison-Wesley, R~ing, MA, 1977. merits, nor on the spelling of "data base". J. Martin, Computer Data-Base Organization, 2n~--e~., Prent~'~6~Ha~[ The mention or lack o~ mention of specific Eng]ewood Cliffs, NJ, 1977. commercial products does not imply endorsement or disapproval by the National I thank John Berg, Joe Collica, Don Bureau of Standards. Deutsch, Dennis Fife, Liz Fong, Len Gallagher, Alan Goldfine, Terry Hardgrave, The reader should have some Familiarity Be1 Leong-Hong, Chuck Sheppard, and Ben with data base management and with the two Shneiderman, whose libraries I raided models. For background on the relational innumerable times while amassing this model, see sources such as: bibliography. Terry Hardgrave and Matt Koll reviewed a draft of this paper. D. D. Chamberlin, "Relational Several authors helped me obtain Data-Base Management Systems", publications. CONTRIBUTION OF THE NATIONAL BUREAU OF The annotated bibliography follows: Sm~NDARDS. NOT SUBJECT TO COPYRIGHT. 55 - M. Adiba and C. De]obel, "The Problem of system which provides a high level the Cooperation Between Different relational data interface". [One D.B.M.S.", G. M. Nijssen (Ed.~ , section of the paper explains that the Architecture and Models in Data Base interface] "is designed in such a way Management- Systems (Proc. IFIP TC-2 that programs can be written on top of Wor~~---Con~ on Modelling in Data Base it to simulate 'navigation oriented' Management Syst., Jan. 1977), database interfaces". "In general our North-Holland, Amsterdam, Neth., 1977 , pp. strategy will be to represent each 1£5-1~6. record type as a relation and to represent information about ordering "We are investigating here the problem and connections between records in the of the cooperation between several form of explicit fields in the DBM~. This cooperation can take place, corresponding relations". [Another for example, via a computer network. section states that an] "important we address ourselves especially to the access path is a binary link". "Binary description of a globa] view of several links are similar to the notion of an data bases". "We have proposed some owner coupled set with manual mechanisms to transform any program membership found in the DBTG written in terms of the global view in specifications". "The main use of programs which can be executed at the binary links in System R is to connect local data base level". [See also child tuples to a parent based on value (Adiba, De]obeY, and L6onard, 1976) and matches in one or more fields". [See (Aaiba and Portal, Igv~)J. also (Lorie and Ni]sson, 1979)]. M. Adiba, C. De]obe], and M. L6onard, "A Auerbach Information Management Series on Unified Approach for Modelling Data in Data Base Management, "A Taxonomy of Data Logical Data Base Design", G. M. Nijssen Structure Models", written by G. C. (Ed.) , Modelling in Data Base Management Everest, Portfolio 21-03-~I, Auerbach Systems (Proc. TFIP TC-2 Wor~g Conf. on Publishers, Pennsauken, NJ, 1979. Modelling in Data Base Management Syst., ]an. ]97~) , North-Holland, Amsterdam, "This Portfolio presents a more Neth., Iq7~, pp. 311-~38. comprehensive taxonomy of data structures; it describes the basic "Here we examine how ... conversion can types of structures, highlighting their be made (in both ways) between network major characteristics and contrasting and relational concepts, using the them through the use of a common DBTG-CODASML and SOCRATE systems as example". [See also (Everest, 1976)]. examples". [See also (Adiba and Delobel, 1977~ and (Adiba and Portal, Igv~)]. C. W. Bachman, "The Data Structure Set Model", in (Rustin, 1974), pp. 1-10. M. Adiba and D. Portal, "A Cooperation "The relational debate, as I see it, is System For Heterogeneous Data Base over style in retrieval languages, and Management Systems", Information Syst., style, really, is only one aspect of a Pergamon Press, Oxford, UK, vol. 3, no. 3, data base system". "I accept the 1978, Pp. ~9-215. relational view and the data-structure-set view as being "We propose here a functional fundamentally compatible as they are architecture of a cooperation system applied in practice, and I'd like to For heterogeneous DBMS in a network demonstrate the equivalence between the environment" [The paper assumes] "the two views". existence of different local data bases already in use under DBMSs like IMS, Codasyl-]ike systems, SOCRATE, C. W. Bachman, "The Role Data Model re]ationa] systems, etc.". "The Approach to Data Structures", Proc. Intl. cooperation system should be able ... Conf. on Data Bases, Heyden, Lon~, UK, to provide facilities for creating and J~y 198~, pp. 1-18. modifying a global view of these local data bases". [See also (Adiba and "This paper examines the Role data Delobel, ]977) and (Adiba, Delobe], and model as an evolutionary step forward L~onard, 19v6)]. in the sequence of data models" [and compares it with other data models]. M. M. Astrahan et a]., "System R: Relationa] Approach to Database C. W. Bachman, "Trends in Database Management", ACM Trans. on Database Syst., Management -- 1975", Proc. Natl. Computer vo]. 1, no. ~, June ~h"~6, pp. 9~-137. Conf., AFIPS, vol. 44, May 1975, pp. ~-576. "System R is a database management - 56 - [One section says] "I consider the data transfer, and a global view of relational model and the different data bases in a distributed data-structure-set model essentially system. The author states that with compatible ann subiect to some restrictions upon the relational transformation From one form to the and CODASYL models], "the 'expressive other". "The relational mode] and the" powers' of the models are equivalent". [data independent accessing mode]] [See also (Borkin, 1978)]. "appear to be completely at odds with each other" "By comparison, the data-structure-set mode] seems to be an J. Bradley, "An Extended Owner-Coupled Set effective hybrid between these two Data Model and Predicate Calculus for extremes". Database Management", ACM Trans. on Database Syst., vol. 3, no. 4, Dec. 197~, pp. 385-416. R. Bell and P. M. D. Gray, "Description of Access Paths for Realising Relations from [In this data model,] "the logical view a CODASYL Database", Research Rep. of the data in the database is AUCS/TR-R~I, Dept. of Computing Sci., U. essentially relational but with the of Aberdeen, Aberdeen, UK, Jan. ]98~. superimposition of the owner-coupled set concept". "Extended Bachman "The ... system generates Fortran diagrams in modified form can easily be programs to retrieve data from a used in the design of relational CODASYL database in the form of databases". relations". "The system generates an internal description of a relation called a 'traversa]', which includes A. F. Cardenas and M. H. Pirahesh, "Data elements representing both algebraic Base Communication in a Heterogeneous Data operators and CODASYL DML commands". Base Management System Network", Information Syst., Pergamon Press, Oxford, UK, "~'?--~, no. l, 1980, pp. 55-79. H. Biller, "On the Equivalence of Data Base Schemas -- A Semantic Approach to "An architectural approach is outlined Data Translation", Information Syst., ... in which any user in any network Pergamon Press, Oxford, UK, v ol. a, no. I, node can be given an integrated and Iq79, pp. RS-4V. tailored view or schema (e.g. hierarchical, relational), while in "The equivalence of data base states reality the data may reside in one and data base schemas is defined". "It single data base or in physically is shown, how a semantic data model and separated data bases, managed data definition language (LDDL) can be individually by the same type of GDBMS used to construct a correct (e.g. CODASYL, IMS, relational) or by specification of a translation different GDBMS".