The Essence of Databases, by F.D. Rolland, is designed to be an undergraduate text or part of a module of a graduate course. It is part of an "Essence of Computing" series by Prentice-Hall. Titles on Program Design and C++ are available and a title on Compilers is promised. Reading this book makes me curious about the others.
The Essence of Databases contains chapters on database modeling and design, relational concepts, SQL query and data definition commands, hierarchic and network data manipulation. Object-oriented databases (with an example in C++) make an appearance in Chapter Seven.
There is also some coverage of database tuning and a stab at database administration. Distributed databases are not given enough coverage to be useful if this is your primary interest.
I found the book to be tightly organized and tightly written. In addition to the intended audience, it may be of interest to someone in the trenches who has wandered or been pulled into databases; perhaps by learning a database package by osmosis.
What would a working instructor say about this book? My feeling is that given the number of pages, Rolland has done a good job in developing succinctly the expected "hit list" items in each area, without sacrificing what a mathematician would call the rigor of the discussion.
One area with intrinsic rigor is the relational database model, which has a firm grounding in the mathematical theory of relations. The coverage in this book is heavily balanced in favor of relational databases. Rolland asserts that the relational model has been the most successful in industry, while also averring that object-oriented databases have been given the most research attention in recent years.
Because of the relational model's grounding in mathematics, one can establish whether a particular database is "fully-relational" or not. I was not at all surprised to learn from this book that not all industry database products are created equal in those terms. Indeed, I was put in mind of the year 1991, when "object-oriented" was on the cover of every computing magazine and in every advertisement, and you had to wonder -- was it marketing's version of OO, engineering's version, or something else entirely.
What did I learn?
- Traversing a database using a network approach is tedious and reminiscent of jumping from node to node through a tree or graph in your first Pascal course.
- A database design is not the same as the "physical level" of the database, although some people speak as if it were.
- DBMSs were a bold advance over storing data in a group of related files; letting various applications be written on top of the same underlying data; all the while transparently managing the physical translation to the underlying in physical storage.
- Relational theory provides a strict way to get your database into higher and higher normal forms, but in practice it isn't always necessary or desirable to have your database in the highest NF possible.
I dimly recall that I may have heard all of this before, but it was enjoyable revisiting the territory under Rolland's terse tutelage.
— Electronic Review of Computer Books