Anson the Gnome

I made a brief mention of The Art of SQL, but in light of my last fairly 'meh' book review, I wanted to talk about good books a bit more. I've been working with databases for a few years now, and I had some really great mentors on the subject early on... so I have felt pretty comfortable with my level of expertise in the area. Still, Faroult and Robson completely blew me away with this book, and reading it helped me understand and solve a whole new range of problems in databases.

The attempt at mapping chapters and subjects onto The Art of War is a bit of a waste of time, but it seems to mostly involve goofy chapter titles and preface. More frustrating for me personally was the discussion of (yes, here we go again) trees and hierarchies in SQL. It's a nice discussion of accessing data in hierarchical formats, but it avoids handling some very important areas: nodes with multiple parents (consider MMO crafting recipes as an example: ingredients can be used in multiple recipes, and some recipes yield things that can be used as ingredients in other recipes), and on-the-fly inserts into trees.

The authors' best advice regarding trees, incidentally, is "SQL still lacks a truly adequate, scaleable[sic] processing of tree structures." It's hard to disagree with that conclusion, and it's hard to argue the authors' decision - in light of that conclusion - to not fully address tree structures in SQL. Out of curiosity I'll probably take a look at Joe Celko's Trees and Hierarchies in SQL for Smarties soon, but to be honest I think professionally my solution is going to be "stay the hell away."

However, in general the book covers a lot of ground in much better depth than most books - offering interesting examples, SQL demonstrations of principles, and performance comparisons of different approaches that is well worth reading, and keeping handy as a reference. It is, frankly, the only database book I've read thoroughly and still been satisfied enough with to recommend.

Agile Database Techniques was a bit of a disappointment. In particular, it was targeted at getting DBAs to be more 'agile' and open to an iterative, evolving database design... which I can agree with, but I didn't need a book for that. Past that, the book spends a lot of time introducing the reader - not "agile software developers" as the title implies, but mostly DBAs - to agile strategies, database design, and finally (finally!) object/relational problems and solutions.

To some extent, object/relational mappings are a solved problem. Object Relational Brokers can do all the heavy lifting for you if you're using a language like Java or C# (I'm not), and if you're not worried about the number of database round trips required to gather your data (or update it). Add in a hierarchy of objects containing other objects (in addition to inheritance), and you're basically done with ORBs and/or performance.

That, really, is the bugbear I should have been talking about a couple of years ago. It's not some vague "databases aren't fast enough" problem, it's "collecting a tree of objects, each of which is also in an inheritance tree, is slow." Individual objects are no big deal... definitely a solved problem. A collection of objects in a language with limited run-time reflection, on the other hand...

I had hoped that Agile Database Techniques would address these larger-scale object/relational mapping problems, but unfortunately it has not. It appears very much targeted at the enterprise software market (one where, in my opinion, "agile" is relative), and avoids (predates?) the web-oriented agile development community (where I have also been looking, to some extent).

So, it is a decent read, and puts in one place a lot of information on iterative database development in enough depth that you can connect the dots and do it yourself (e.g., test-driven development is a perfectly valid strategy for database schema design too). Not so awesome if you're not a DBA, not working in enterprise software, or dealing with the particular problems of games, but still a decent read. At $26 from Amazon, it may be worth putting on your bookshelf... as long as your expectations are reasonable. :)

Entries from October 2008

Sunday, October 19. 2008

The Art of SQL

Wednesday, October 1. 2008

Agile Database Techniques: Effective Strategies for the Agile Software Developer

Subscribe to my newsletter

Tags

Blogroll