History of Database Management Systems
We have already seen that the predecessor to the DBMS was the file-based system.
However, there was never a time when the database approach began and the file-based
system ceased. In fact, the file-based system still exists in specific areas. It has been
suggested that the DBMS has its roots in the 1960s Apollo moon-landing project, which
was initiated in response to President Kennedy’s objective of landing a man on the moon
by the end of that decade. At that time there was no system available that would be able
to handle and manage the vast amounts of information that the project would generate.
As a result, North American Aviation (NAA, now Rockwell International), the prime
contractor for the project, developed software known as GUAM (Generalized Update
Access Method). GUAM was based on the concept that smaller components come
together as parts of larger components, and so on, until the final product is assembled. This
structure, which conforms to an upside-down tree, is also known as a hierarchical structure.
In the mid-1960s, IBM joined NAA to develop GUAM into what is now known
as IMS (Information Management System). The reason why IBM restricted IMS to the
management of hierarchies of records was to allow the use of serial storage devices, most
notably magnetic tape, which was a market requirement at that time. This restriction was
subsequently dropped. Although one of the earliest commercial DBMSs, IMS is still the
main hierarchical DBMS used by most large mainframe installations.
In the mid-1960s, another significant development was the emergence of IDS (Integrated
Data Store) from General Electric. This work was headed by one of the early pioneers of
database systems, Charles Bachmann. This development led to a new type of database system
known as the network DBMS, which had a profound effect on the information systems
of that generation. The network database was developed partly to address the need to
represent more complex data relationships than could be modeled with hierarchical structures,
and partly to impose a database standard. To help establish such standards, the Conference
on Data Systems Languages (CODASYL), comprising representatives of the US
government and the world of business and commerce, formed a List Processing Task Force
in 1965, subsequently renamed the Data Base Task Group (DBTG) in 1967. The terms
of reference for the DBTG were to define standard specifications for an environment that
would allow database creation and data manipulation. A draft report was issued in 1969
and the first definitive report in 1971. The DBTG proposal identified three components:
n the network schema – the logical organization of the entire database as seen by the
DBA – which includes a definition of the database name, the type of each record, and
the components of each record type;
n the subschema – the part of the database as seen by the user or application program;
n a data management language to define the data characteristics and the data structure, and
to manipulate the data.
For standardization, the DBTG specified three distinct languages:
n a schema Data Definition Language (DDL), which enables the DBA to define the
schema;
n a subschema DDL, which allows the application programs to define the parts of the
database they require;
n a Data Manipulation Language (DML), to manipulate the data.
Although the report was not formally adopted by the American National Standards
Institute (ANSI), a number of systems were subsequently developed following the
DBTG proposal. These systems are now known as CODASYL or DBTG systems. The
CODASYL and hierarchical approaches represented the first-generation of DBMSs.
We look more closely at these systems on the Web site for this book (see Preface for the
URL). However, these two models have some fundamental disadvantages:
n complex programs have to be written to answer even simple queries based on navigational
record-oriented access;
n there is minimal data independence;
n there is no widely accepted theoretical foundation.
In 1970 E. F. Codd of the IBM Research Laboratory produced his highly influential paper
on the relational data model. This paper was very timely and addressed the disadvantages
of the former approaches. Many experimental relational DBMSs were implemented
thereafter, with the first commercial products appearing in the late 1970s and early 1980s.
Of particular note is the System R project at IBM’s San José Research Laboratory in
California, which was developed during the late 1970s (Astrahan et al., 1976). This
project was designed to prove the practicality of the relational model by providing an
implementation of its data structures and operations, and led to two major developments:
n the development of a structured query language called SQL, which has since become
the standard language for relational DBMSs;
n the production of various commercial relational DBMS products during the 1980s, for
example DB2 and SQL/DS from IBM and Oracle from Oracle Corporation.
Now there are several hundred relational DBMSs for both mainframe and PC environments,
though many are stretching the definition of the relational model. Other examples
of multi-user relational DBMSs are Advantage Ingres Enterprise Relational Database from
Computer Associates, and Informix from IBM. Examples of PC-based relational DBMSs
are Office Access and Visual FoxPro from Microsoft, InterBase and JDataStore from
Borland, and R:Base from R:Base Technologies. Relational DBMSs are referred to as
second-generation DBMSs. We discuss the relational data model in Chapter 3.
The relational model is not without its failings, and in particular its limited modeling
capabilities. There has been much research since then attempting to address this problem.
In 1976, Chen presented the Entity–Relationship model, which is now a widely accepted
technique for database design and the basis for the methodology presented in Chapters 15
and 16 of this book. In 1979, Codd himself attempted to address some of the failings in his
original work with an extended version of the relational model called RM/T (1979) and
subsequently RM/V2 (1990). The attempts to provide a data model that represents the ‘real
world’ more closely have been loosely classified as semantic data modeling.
In response to the increasing complexity of database applications, two ‘new’ systems
have emerged: the Object-Oriented DBMS (OODBMS) and the Object-Relational
DBMS (ORDBMS). However, unlike previous models, the actual composition of these
models is not clear. This evolution represents third-generation DBMSs, which we discuss
in Chapters 25–28.
We have already seen that the predecessor to the DBMS was the file-based system.
However, there was never a time when the database approach began and the file-based
system ceased. In fact, the file-based system still exists in specific areas. It has been
suggested that the DBMS has its roots in the 1960s Apollo moon-landing project, which
was initiated in response to President Kennedy’s objective of landing a man on the moon
by the end of that decade. At that time there was no system available that would be able
to handle and manage the vast amounts of information that the project would generate.
As a result, North American Aviation (NAA, now Rockwell International), the prime
contractor for the project, developed software known as GUAM (Generalized Update
Access Method). GUAM was based on the concept that smaller components come
together as parts of larger components, and so on, until the final product is assembled. This
structure, which conforms to an upside-down tree, is also known as a hierarchical structure.
In the mid-1960s, IBM joined NAA to develop GUAM into what is now known
as IMS (Information Management System). The reason why IBM restricted IMS to the
management of hierarchies of records was to allow the use of serial storage devices, most
notably magnetic tape, which was a market requirement at that time. This restriction was
subsequently dropped. Although one of the earliest commercial DBMSs, IMS is still the
main hierarchical DBMS used by most large mainframe installations.
In the mid-1960s, another significant development was the emergence of IDS (Integrated
Data Store) from General Electric. This work was headed by one of the early pioneers of
database systems, Charles Bachmann. This development led to a new type of database system
known as the network DBMS, which had a profound effect on the information systems
of that generation. The network database was developed partly to address the need to
represent more complex data relationships than could be modeled with hierarchical structures,
and partly to impose a database standard. To help establish such standards, the Conference
on Data Systems Languages (CODASYL), comprising representatives of the US
government and the world of business and commerce, formed a List Processing Task Force
in 1965, subsequently renamed the Data Base Task Group (DBTG) in 1967. The terms
of reference for the DBTG were to define standard specifications for an environment that
would allow database creation and data manipulation. A draft report was issued in 1969
and the first definitive report in 1971. The DBTG proposal identified three components:
n the network schema – the logical organization of the entire database as seen by the
DBA – which includes a definition of the database name, the type of each record, and
the components of each record type;
n the subschema – the part of the database as seen by the user or application program;
n a data management language to define the data characteristics and the data structure, and
to manipulate the data.
For standardization, the DBTG specified three distinct languages:
n a schema Data Definition Language (DDL), which enables the DBA to define the
schema;
n a subschema DDL, which allows the application programs to define the parts of the
database they require;
n a Data Manipulation Language (DML), to manipulate the data.
Although the report was not formally adopted by the American National Standards
Institute (ANSI), a number of systems were subsequently developed following the
DBTG proposal. These systems are now known as CODASYL or DBTG systems. The
CODASYL and hierarchical approaches represented the first-generation of DBMSs.
We look more closely at these systems on the Web site for this book (see Preface for the
URL). However, these two models have some fundamental disadvantages:
n complex programs have to be written to answer even simple queries based on navigational
record-oriented access;
n there is minimal data independence;
n there is no widely accepted theoretical foundation.
In 1970 E. F. Codd of the IBM Research Laboratory produced his highly influential paper
on the relational data model. This paper was very timely and addressed the disadvantages
of the former approaches. Many experimental relational DBMSs were implemented
thereafter, with the first commercial products appearing in the late 1970s and early 1980s.
Of particular note is the System R project at IBM’s San José Research Laboratory in
California, which was developed during the late 1970s (Astrahan et al., 1976). This
project was designed to prove the practicality of the relational model by providing an
implementation of its data structures and operations, and led to two major developments:
n the development of a structured query language called SQL, which has since become
the standard language for relational DBMSs;
n the production of various commercial relational DBMS products during the 1980s, for
example DB2 and SQL/DS from IBM and Oracle from Oracle Corporation.
Now there are several hundred relational DBMSs for both mainframe and PC environments,
though many are stretching the definition of the relational model. Other examples
of multi-user relational DBMSs are Advantage Ingres Enterprise Relational Database from
Computer Associates, and Informix from IBM. Examples of PC-based relational DBMSs
are Office Access and Visual FoxPro from Microsoft, InterBase and JDataStore from
Borland, and R:Base from R:Base Technologies. Relational DBMSs are referred to as
second-generation DBMSs. We discuss the relational data model in Chapter 3.
The relational model is not without its failings, and in particular its limited modeling
capabilities. There has been much research since then attempting to address this problem.
In 1976, Chen presented the Entity–Relationship model, which is now a widely accepted
technique for database design and the basis for the methodology presented in Chapters 15
and 16 of this book. In 1979, Codd himself attempted to address some of the failings in his
original work with an extended version of the relational model called RM/T (1979) and
subsequently RM/V2 (1990). The attempts to provide a data model that represents the ‘real
world’ more closely have been loosely classified as semantic data modeling.
In response to the increasing complexity of database applications, two ‘new’ systems
have emerged: the Object-Oriented DBMS (OODBMS) and the Object-Relational
DBMS (ORDBMS). However, unlike previous models, the actual composition of these
models is not clear. This evolution represents third-generation DBMSs, which we discuss
in Chapters 25–28.
No comments: