At last! A book that tells it like it is!
In the age of The Object, many unenlightened I.T. professionals dismiss the database as a mere "persistent object repository". In other words, a big bucket to put data into. At best this approach results in poor performance - at worse, you're looking at project failure, loss of reputation, dismissal, homelessness, etc.
Jonathan Lewis's book is not a magic bullet to solve all Oracle project ills nor is it an easy read. The first chapter alone covers some difficult ground regarding redo and rollback. What becomes immediately apparent is that databases are extremely complex beasts. However, when you put the effort in, take heed of good advice, and, most importantly, prove it to yourself, there is a much greater chance of going live with a champ rather than a chump!
If you can get through the (essential) first four chapters unscathed, the rest of the book is plain sailing and, to a certain degree, you can dip into them. However, I wouldn't say that the book contains "tips": It contains "pointers to the right direction and away from the wrong turns". Quite often it is tempting to use a feature of Oracle "because it's there". However, Mr. Lewis tells us whether its worth it of not - preventing much potential time-wasting. Also, to his credit, he isn't afraid to speak out: defending RAID 5, debunking Oracle's Objects, and putting RMAN into its proper perspective.
No project involving Oracle should be started without this book being read and understood by all it's key players!