x Preface
Part A establishes the need for data analysis; it differs from the rest of the book in its
'case history' style. The reader who is only too familiar with the problems introduced
here may wish to skip quickly through this part.
Part 1 examines the concepts of database and database management system (DBMS).
A DBMS is viewed not so much as a good thing in its own right, but more as a means of
overcoming the problems of shared data. The architecture of a DBMS is discussed in
terms of the ANSI/SPARC model.
Part 2 explains how a data model can be designed in a 'bottom-up' direction by
applying normalisation techniques to the data items (attributes). The usual way of
treating normalisation is to beat a trail down the first, second, third, fourth, and maybe
fifth, normal form route, with Boyce/Codd normal form snapping hard at the heels of
third normal form. I cannot see the merit of this approach, which seems to have become
prevalent for historical reasons rather than for its intrinsic worth, and I have a sneaking
feeling that it is sustained chiefly by its convenience as a source of examination
questions. My approach has been to distinguish carefully between duplicated data and
redundant data, and hence establish the Boyce/Codd rule directly. Tal-.les (i.e.
relations) that satisfy the Boyce/Codd rule are described as weU-normalised. Ev,.n wellnormalised
tables may still contain redundant data; once this is eliminated the tables
are said to be fully-normalised. Because the Boyce/Codd rule is stated in terms of the
concept of a determinant, I often use the terms determines and determinancy where it
might be more usual to find the opposite viewpoint being taken, coupled with the use of
terms such as is dependent on and dependency. I have not dealt with fifth normal form
explicitly as, although theoretically interesting, it is of little practical significance.
Part 3 builds on the results of Part 2, but takes an opposite 'top-down' approach
which starts by identifying entity and relationship types, and then uses these to
construct a framework into which the attributes may be slotted. Part 3 concludes with a
discussion of how the design may be 'flexed' to improve its performance.
Part 4 covers several topics related to data analysis. The concepts of Part 3 are applied
to distributed database systems. Data manipulation is discussed in terms of relational
algebra, which is then applied to optimisation issues. The SQL language is introduced
(based on SQL92). Object-orientation and object-relational developments (based on
SQL99) are considered. The Codasyl (network) model is described in an appendix, as a
contrast to the relational model.
x Preface
Part A establishes the need for data analysis; it differs from the rest of the book in its
'case history' style. The reader who is only too familiar with the problems introduced
here may wish to skip quickly through this part.
Part 1 examines the concepts of database and database management system (DBMS).
A DBMS is viewed not so much as a good thing in its own right, but more as a means of
overcoming the problems of shared data. The architecture of a DBMS is discussed in
terms of the ANSI/SPARC model.
Part 2 explains how a data model can be designed in a 'bottom-up' direction by
applying normalisation techniques to the data items (attributes). The usual way of
treating normalisation is to beat a trail down the first, second, third, fourth, and maybe
fifth, normal form route, with Boyce/Codd normal form snapping hard at the heels of
third normal form. I cannot see the merit of this approach, which seems to have become
prevalent for historical reasons rather than for its intrinsic worth, and I have a sneaking
feeling that it is sustained chiefly by its convenience as a source of examination
questions. My approach has been to distinguish carefully between duplicated data and
redundant data, and hence establish the Boyce/Codd rule directly. Tal-.les (i.e.
relations) that satisfy the Boyce/Codd rule are described as weU-normalised. Ev,.n wellnormalised
tables may still contain redundant data; once this is eliminated the tables
are said to be fully-normalised. Because the Boyce/Codd rule is stated in terms of the
concept of a determinant, I often use the terms determines and determinancy where it
might be more usual to find the opposite viewpoint being taken, coupled with the use of
terms such as is dependent on and dependency. I have not dealt with fifth normal form
explicitly as, although theoretically interesting, it is of little practical significance.
Part 3 builds on the results of Part 2, but takes an opposite 'top-down' approach
which starts by identifying entity and relationship types, and then uses these to
construct a framework into which the attributes may be slotted. Part 3 concludes with a
discussion of how the design may be 'flexed' to improve its performance.
Part 4 covers several topics related to data analysis. The concepts of Part 3 are applied
to distributed database systems. Data manipulation is discussed in terms of relational
algebra, which is then applied to optimisation issues. The SQL language is introduced
(based on SQL92). Object-orientation and object-relational developments (based on
SQL99) are considered. The Codasyl (network) model is described in an appendix, as a
contrast to the relational model.
การแปล กรุณารอสักครู่..