Customer Reviews

Mastering Data Modeling: A User-Driven Approach

5 star:		(6)
4 star:		(0)
3 star:		(2)
2 star:		(1)
1 star:		(0)

 

 

I disagree that a person could become a "master" data modeler if the contents of this book are the complete set of skills in their arsenal. While the book outlines some good techniques for interviewing end users and basic data modeling skills, there is alot more involved in data modeling than what is covered here.

As an experienced data modeler who works with large, complex data models in a constantly changing business, I find I do not refer to this book at all. The book excludes common data modeling constructs that I have found very useful, including subtypes and supertypes. The book does not explain the difference between conceptual, logical, and physical data modeling. (It covers techniques used to capture conceptual/logical level data, but nowhere does it explain that or the difference between this type of model and a physical model, and why and when you'd need one or the other.)

The book does not cover normalization, which, once one leaves the interview with end users, one will need to understand. The book does not mention data integration with other systems or databases, how this topic is important and could (and often should) arise in interviews with end users.

Some of the topics covered I found shallow and incomplete, for example, how to name things in a data model. The authors take a parochial view by ignoring real world issues such as using consistent names across database and organizations, and avoiding naming things for what they are used for, not what they are.

As a practicing data modeler, I find my users aren't as naieve about data models as Carlis and Maguire assume them to be. I often am asked why I am modeling data in a given way. In my view, this book does not address the "why" - why do you model the data in the way suggested, and what happens if you don't. When I can answer these questions well for my customers, I earn approval, and this book doesn't equip one to do so.

In sum, my belief is that this book contains about 1/4 of the information a person needs to know to become a "master" data modeler. It's a good starter book if you are a novice data modeler or are having trouble gathering information from business subject matter experts, but if you really want to become an expert data modeler, I'd recommend continuing beyong this book. I prefer 'Data Modeling Essentials 2nd Edition' by Graeme Simsion

The book describes a method to structure any given sets of data according to generic rules. Eventhough my background does not allow me to judge the theoretical validity of the method, the book is easy to read and all the concepts are easy to understand and described in details. I have applied the Carlis and Maguire method for modeling data in a small research group and it is brilliant. The method allows users to discuss their data in their own language and the modeler can build a logical representation which is understood and well accepted by the users. I will certainly use this book and the method for any future database design.

The secret is out!

I've been using the techniques described in this book for years because one of the authors taught me. I've used them to model data about research science, business, and topology. Now others can learn it too.

Carlis cured me of normalization. There's a difference between normalization and "normal forms". A goal of modeling is to produce databases in high normal forms - Boyce-Codd Normal Form, fifth normal form, etc... Most modelers think the only way to do this is through normalization, a specific process that step-by-step improves to a draft model. This book shows how to avoid that process completely. I used to do normalization. Now I use the conversational techniques of this book to reach high normal forms sooner. One thing I always hated about Normalization was that I usually did it after talking to users, which means I was making decisions that the users should have been making. I have not performed normalization in at least ten years. Yet I still produce databases in high normal form. This book does include a chapter about normalization, with normal forms up to fifth, so you can see for yourself how the technique produces high-normal-form databases. If you learned that normalization was essential part of data modeling, this chapter will help you learn this different way of working. If you are new to data modeling, you should start with this book to avoid learning normalization altogether. The principles of high normal forms are important, but the process of normalization is ludicrous.

This is a book about data modeling, not physical database design. It concentrates on the modeling in users' language. The naming conventions it recommends are based on guidelines of language and categories. If you follow these naming guidelines, you will not need to learn a huge list of more specific, special-case naming rules.

I also like what Carlis and MaGuire say about constraints. By following their constraint advice, I have become a much faster data modeler, and my team mates (programmers, DBAs) do not have to wait so long for me to finish my work. It also helps me keep my data models flexible, good for a changing business environment.

This book has more examples than any book on modeling I have ever seen. I stopped counting sample data models when I got to 300.

The hardest part of application design is understanding the user's data. This book concentrates on solving that problem, leaving the technical details of database design to other books.

This book will be on my table always. It cuts through the computer science's obsession for esoteric notations and undue rigor (that scares the end users even before the analyst has had a chance to begin!) and puts the user needs at their right place: right in the center. I have used this methed several times now with exciting results. Users are more forthcoming, there are lesser I-thought-this-when-you-said-that instances. Two thumbs up for the excellent work!!!

Messrs. Carlis and Maguire take the long overdue postition that data models should be tools for communicating an analyst's understanding of a business with the people in that business BEFORE communicating with technical designers. This means that models should not be overly complex, should be described in plain language, and should be organized for presentation. Their notation is a simplified version of information engineering that removes pieces that are confusing and actually of no consequence. It is a refreshing change from the likes of the UML and IDEF1x.

I just finished a semester-long graduate course that used this book as one of three textbooks. I am an experienced, mid-career data modeler and developer of database-centered applications and I have used entity-relationship diagrams for years, so I was initially a little concerned that this text would not have much new for me. I was also suspicious of the author's notation because it leaves out a number of syntactical elements I was accustomed to using in UML and other data modeling languages. However, I was very pleased and surprised with the amount of valuable and original content in this book, and I have significantly altered my modeling practices as a result.

This book is written in a quirky but very readable style, and I recommend it for both experienced data modelers who want to improve the quality and efficiency of their work, and for anyone who wants to start learning about data modeling.

Beginning data modelers will especially benefit from
1) the emphasis on logical data modeling and user communication,
2) the explanation of how to choose good names and identifiers, and
3) the examination of common patterns that appear in almost every data model, how to recognize them, and how to interpret them.

Experienced data modelers who have already grappled with those issues on past projects will gain further insight from the authors' coherent systemization of those topics, but should also appreciate and benefit from
1) the content-neutral process laid out for iteratively evolving a diagram through controlled interactions with users,
2) the investigation of the underlying principles and forces that shape a data model, with many illustrating examples drawn from real-world projects, and
3) the intellectually rigorous analysis of the notation and syntax used to represent a data model, and the semantic interpretation of each structure.

If I had to identify one set of people who would get the most out of this book, I would select software developers who learned data modeling on the job, or who are starting to learn it, and who would like a cohesive intellectual foundation for what they are doing. This book can help you go from being a "pretty good" data modeler who can handle modeling familiar data to being an expert data modeler with a strong theoretical basis for your work who can quickly develop high-quality data models regardless of your personal level of understanding of the underlying subject area.

This book was required reading for a class, or I would not have purchased it.

The content itself was OK, but it spent too much time going into detail on areas that were obvious, and then introduced new topics with a casual comment and then never explained those any further.

On a more practical note, the book was bound on the wrong side, with the cover upside down. You had to hold the book upside down, and then read from the back to the front. I would have had it replaced, however being required for a class I did not have the time to exchange it. On one occasion I actually had a teacher for small children comment on my holding the book upside down. On other occsions small children looked at me very strangely when I was reading.

This book describes how to perform data modeling. It will help you a lot if you are working as a system analyst. It tells you how to get information from the users. Also it discusses the data modeling in deep and teaches you how to identify common mistakes very early.
Though the good is very good for modeling, $35 is somehow high for this book.
Another point is that is uses a diagram called LDS. This type of diagram is not well known. You'll suffer to find tools that draw this kind of diagrams and to find other professionals who use it.

I am lucky to have had the opportunity to be a student of John Carlis, so the readability and quality of this book was no surprise. Anyone who has even discussed computer science with him would assure you that his focus is not simply on amassing tools, but on mastering them. This book was loaned to me on the job, and served as an excellent introduction, and as a wonderful reference.

The simplicity and power of LDS makes it an excellent tool for software developers. In a discussion with Carlis, he explained to me the appeal that LDS had to him. He said its simple, limited set of components allows you to use data modeling not as a reference to understand software, but as a way to plan it, and think about the consequences of each decision you make.

Buy this because it's a great read, a great introduction, and an invaluable reference. No software developer's desk should be without it.