Customer Reviews

Practical Issues in Database Management: A Reference for the Thinking Practitioner

5 star:		(12)
4 star:		(1)
3 star:		(2)
2 star:		(1)
1 star:		(2)

 

 

Contains discussions on widely misunderstood, but vitally important, database issues. Considers how things should be according to relational theory, what goes wrong in practice due to failing to understand these fundamental concepts, and provides practical recommendations/workarounds where possible. A bridge between theory and practice. A concise and straightforward read, although to gain maximum benefit you should already have a reasonable understanding of relational databases.

Some of the main points are as follows. Relational databases can support data types of arbitrary complexity ("objects") - but we need DBMS vendors to implement the means to provide such support. Use declarative integrity constraints when possible. Avoid tables that allow duplicate rows - ensure all tables have a primary key (may require a surrogate key). Avoid redundant data. Ensure your database is normalized - avoids numerous problems. Resist the temptation to denormalize. Recognise entity supertypes/subtypes and how to implement as tables. Minimize use of nulls whenever possible. Also discusses climbing trees and quota queries. Separate chapters discuss all these points in detail.

A constant theme throughout the book is how poorly SQL and commercial DBMSs support relational concepts, and the numerous problems this causes. Personally, I believe there is little that most application/database developers can do about this, other than suffer what gets "inflicted" upon us. Pascal urges us to pressurise DBMS vendors to provide better relational support.

This book is firmly of the view that "relational is right". If you agree, you will probably like this book. Most of the references are to C.J Date's writings. However, I suspect some people won't like this book - those who believe "relational is wrong" (e.g. some OODBMS proponents); and those who prefer a "cookbook approach" for their specific DBMS and don't really care about what the "right way" is. I suspect some people will think that Pascal should "live in the real world" rather than worrying about theory, yet this book argues that it is because of the failure to understand fundamental database concepts that leads to "real world" databases having the problems that they do.

Helps database professionals educate themselves in fundamental database concepts, and illustrates the correct way of doing things in relational databases. Well worth reading.

I came to this book with high hopes after having read much of Fabian Pascal's writing on the Web, which is very specific in calling out the many weaknesses of contemporary database practice and much less specific when it comes to strategies for dealing with those weaknesses. Unfortunately, the writing in this book is much like his writings online.

Pascal is clearly a seasoned professional and also someone who understands the relational model well. However, his frustration with the current state of the art bleeds over into this book, and is unhelpful. I now very clearly understand that SQL is incapable of dealing with a truly relational view of data, and that SQL (and by extension almost all contemporary DBMSes) is flawed and illogical in many ways, but there's only so much that I, as a database designer and software developer, can do with that information. In many cases, only a change in phrasing would have been necessary: I think the book would have been much more useful if it had been written prescriptively, as a "style guide" or best practices manual (DON'T use NULL. DO normalize your relations to at least 3NF) with his impassioned critiques of current technology provided as rationales for his guidelines.

Also, there's more than a whiff of the amateur to the presentation in this book. Aside from the baffling reprints of web pages included as appendices to the first chapter, he recycles exhortations across chapters and reuses the same examples over and over (down to the same explanatory text), without necessarily explaining what's different about this usage from all the others. Add to that a large, wide text face and the book, not particularly long at 256 pages, starts to feel significantly padded.

Most interested parties would be much better served by C.J. Date's excellent "Database In Depth". As it is, Pascal cribs heavily from Date's work (2/3 of the book's citations are credited to either Date or his frequent partner Hugh Darwen), and I can't help but feel that Date expresses the same concepts much more clearly and concisely. There's a lot of good material in this book, but it's a frustrating slog to get to it.

This is a decent book to read within the first year of your dealing with relational databases. Most chapters (particularly the ones on normalization and subtypes) are at least in some way informative and will improve your clarity of thought. Whether the book is really practical is open for discussion, since it deliberately stays away from real products, pointing to their deficiencies and suggesting you should demand improvement from the vendors. OK, I'll send letters to Gates and Ellison today. But it _is_ practical in the sense of improving your thinking on the more fundamental level. Which is nothing to sneeze at, of course.

That said, the book has a strong unprofessional taste to it in terms of overall delivery: it's really half a dozen blog entries, padded with endless recaps, intros, and even a strange assemblage of web pages whose connection to the main body of text is tenuous at best (some of them aren't even referred to! Why is this junk inserted in the book? To achieve the obligatory two-hundred-page volume?) A pile of quotes from online posts, attributionless excerpts from 'trade magazines' supposedly illustrating all-pervading imbecility of the database constituency, and so on. The points the author makes are correct, but he belabours them way too much. Where one quote would suffice he gives you five.

A fair number of typos -- nothing terrible, something like the following: say, there's a picture of a table with fields, and then in the text the fields' names are different -- no big deal, but this is a technical book and small discrepancies like that really slow you down. A couple of openly ungrammatical statements (copy editing, anyone?)

Sometimes the book is illogical: for example, there's one place where a line of argument goes like this, if A and [this and that] then A: well, thanks for this refreshing tautology. Huge amount of muddled quotes from self and Date and Darwen and McGovern. Are there any other people worth reading? This smells a bit of self- and mutual promotion, to be honest; like there's this cabal of ever-crusading brothers at arms who keep dropping each others' and no one else's names.

References are very imprecise: mostly something like: "normalization is very important! ([2])". [2] what? Any particular chapter, page?

Quotes are frequently jumbled with the author's own text: you read and suddenly you see that you've been reading an excerpt from another book. Where did it start? Citations are frequently edited to fit the context to the point where square brackets make you dizzy. If a direct quote is so ill-fitting, why not simply paraphrase?

All in all, I agree with Forrest L. Norvell's and Jeff Unsal's reviews below, though unlike that last gentleman, I don't think the book is useless. I think, you probably _should_ read this book, though chances are high that if you've been working with SQL databases for a while you already know all of this from other sources. But maybe not: the author is fundamentally correct when he complains about pervasive vendor hype in the industry (and literature) and widespread ignorance and misunderstaning of fundamentals. So yes, go ahead and read it; at least check it out. Get a discounted copy; be prepared to skip the vapour and chunks of repetitious text, and figure out a number of confusing typos and sloppy narrative, but there *is* value in this book, at least for some readers. Although, if you read a fair amount of Date, you may not need this one -- and of course Date writes much, much better. Otoh (and for what it's worth), the author writes better than Celko.

Anyway, the book is flawed but worthwhile. I would like to see a new edition -- proofread, better structured, with the web pages and other stuffing removed. Maybe even a half of its current size (for my money I prefer to read a lesser amount of clearly written text than twice as much but muddled).

Enjoy.

Ever wondered why is it so difficult to put rich data into the relational model, or why so much of database logic keeps being badly manipulated by application programs instead of by DBMS? Ever felt SQL was too verbose, or complex, or illogical, and simply arbitrarily limited? Mr Pascal helps us organise our grips against current DBMSs and hope for a better future.

This book analyses common database problems every professional with more than a few months in the field has already faced, and relentlessly trace them to failures in SQL implementations or in the SQL language itself. A real mind-opener in the current state of the field, where from press to shop everything is dominated by vendors but few true critics survive.

This may be the most practical database book I will ever read. While it is not wriiten for the complete novice, I would say it is important as a good second book to read (an excellent first book would be Access Database Design and Programming by Steven Roman, or Sams Teach Yourself SQL in 10 Minutes by Ben Forta), after you have developed a rudimentary understanding of SQL. After working with SQL in any DBMS, you will begin to run accross some of the problems he addresses in his book - inconsistent treatment of nulls, duplicate information, etc. Each chapter stands on its own, making it valuable as a reference guide, and in many instances he provides SQL workarounds for problems brought on by...SQL. Although he does lay blame at feet of current DBMS vendors for not fully implementing the relational model, he addresses the fundamental lack of knowledge on the part of practitioners. By far the best chapter is the one on normalization, particularly his discussion on the fourth and fifth normal forms.

If you can not find the gold in a single paragraph, you definitely can find in a single page.

This is the best data modeling book I have ever read. What best I feel that it talks in-depth detail, namely, the problem of data modeling and its solution. In comparison, a lot of data modeling book talks only the rule, simply a clasroom instructor'style.

In fact, data modeling rule is not hard to upderstand, but it is very hard to implement. This book really talks practically.

This book offers clear but challenging thinking about some important issues in relational database design, rules of integrity, and normalization. Pascal emphasizes principles of relational design based on predicate logic and set mathematics in a nice blend of theory and practical hints. He argues that many of the problems in using databases are a result of poor relational design. For example, in discussing duplicates he advocates prohibiting duplicates and the use of keys. Throughout he describes the problems if the data is not normalization.

My only complaint is that he regularly complains about the design of the SQL standard. For example "Due to numerous flaws in the design and implementation of SQL, the performance of SQL DBMS ... leads users to denormalize for performance". As a user I have say over my design, but little sway over the SQL standard or my particular software package. With regard to denormalizing, I think there is also a tendency to use flat tables in web page pages, so perhaps Pascal should go over developing tools to program relational tables easier with ASP pages. I would have also appreciated his thoughts on dimensional modeling used in data warehouses.

This book has been essential in my work as a consultant and developer in the database application field. I use it whenever I need a short, easy to read, and no-nonsense summary of data management fundamentals. This book explains "how it should be", "how it often is", and "what you can do about it", making it a really perfect recommendation in many situations.

I think the above title is what will happen in your career in IT if you read and understand this book. I personally have made a career almost exclusively at rescuing one dysfuntional/unmaintainable IT system after another, simply by going in and applying the principles and understanding outlined in this and Fabian Pascal's earlier book. And I am regularly given "GURU of the GURUs" status for the only reason that I understand and apply these simple principles in IT applications...

Bravo Zulu (well done) to Mr. Pascal. Very respectful to the reader, just as the title implies. Wonderful, delightful reading, amazingly engaging style. If you choose to engage, you will benefit beyond belief. I would buy this book for you if I could. Choose not to engage then you may write a bad review, but you still get the benefit of exposure to it. Given current situation, that exposure is still a step in the right direction. Scarce publication on this subject matter makes it worth it's weight in gold. This is not a "book for dummies" rather a book for smart folk. New words and ideas are introduced to help us understand old concepts. This book is a call for action for our discipline, a challenge for those of us dedicated to make a difference in this world. The revolution is based on this discipline, pay homage to it.