Re: CODASYL-like databases

From: Ken North <>
Date: Thu, 03 Apr 2008 20:41:25 GMT
Message-ID: <VhbJj.21$>

> I agree with this wrt hierarchcial dbms's, not network. In 1970,
> hierarchical dbms products were everywhere, network dbms products
> hardly existed.

Bachman's GE IDS was available in the late '60s and it heavily influenced the network model (CODASYL standard). Charles Bachman was a key player in the task group that released the first database standard in 1971. IBM was originally involved in the DBTG but it put its focus on IMS instead of a CODASYLS DBMS. Other computer companies eventually jumped in with CODASYL-compliant products.

> Before 1970, database management wasn't even a computer science
> concept.

The notion of using a data base, instead of ad hoc data stores, dates back long before 1970. The survey of the CODASYL data base task group in 1968 included dozens of existing products - reproduced here:

Systems such as IBM GIS (1966), GIM (1965) and IDS (1965) were quite sophisticated. For example, GIM implemented demand-paged memory managed by the DBMS executive (because virtual memory wasn't yet an IBM OS feature). It dynamically loaded and cached code pages and data pages during an era when CPU memory was expensive.

This page highlights early contributions to database technology:

> Until at least 1974 ...

In October, 1974 my lab cooperated with UCLA in running the "Comparative Data Base Management Systems Seminar". In two days, we covered the leading 12 DBMS products (with assistance form the vendors). None of those products were based on Codd's relational model and IBM presented IMS. The ACM Computing Surveys of March 1976 (Vol. 8, Number 1) focused on "Data-Base Management Systems. Guest editor E.H. Sibley's introduction was titled "The Development of Data-Base Technology". Sibley wrote:

"Data-base technology is one of the most rapidly growing areas of computer and information science. In less than twenty years, with the greatest part of the development in the past eight years, data-base systems have come from nothing to be a topic of major interest."

So Sibley refers to a period from 1956 on, with the major development on data-base systems occuring from 1968 to 1976.

> 3. > The early RDBMS products were, to some extent, an attempt to
> obtain the power and simplicity of the relational model without
> discarding a code base that had been based on hierarchical or network
> models of databases.

I'm not sure I understand what code base was not discarded.

The code to operate on hierarchical and network databases was primarily in programs written in FORTRAN, COBOL and PL/I. And some DBMS products of that era had only an integrated (interpreted) query and data manipulation language, without an interface for higher-order (compiled) languages.

Products such as Ingres, Oracle and Informix offered embedded SQL and pre-compilers for languages such as COBOL and C. However, the logic of CODASYL database applications is quite different from SQL applications. It's not the path of least resistance to try to create SQL applications by starting with the code of a CODASYL or hierarchical database application.

The SQL DBMS products may have used some common OS code, such as buffer managers and interrupt service routines, but preserving the investment in older database software wasn't generally a goal for database companies pushing the relational model. Received on Thu Apr 03 2008 - 22:41:25 CEST

Original text of this message