Expert Oracle Exadata [Paperback]

This is a fantastic technical tome on all things Exadata. Outstanding coverage includes intro of the technology, hardware configurations, administration tasks, performance monitoring and diagnostics, wait events, backup & recovery, and application SQL tuning. Quite easily one of the two best books I've ever read regarding any Oracle technology (The other being Guy Harrison's Oracle Performance Survival Guide). I've met some of the book's authors over the years at various Oracle events (e.g. Hotsos) and thus knew these guys were top notch. But this book still totally blew me away even expecting it to be great knowing who it was coming from. I've been doing Oracle databases for over 20 years, written nine book myself, taught for Oracle Education, and worked extensively with all things Oracle. In fact I'm an Oracle ACE. So I know a good book when I see one - and this book is tops. You will not be disappointed - worth every penny.

This book is completely different from the other book on EXADATA (the one by Oracle Press and the Oracle Product Managers). The other books is not bad, just for a different audience. Expert Oracle EXADATA is very practitioner oriented.

The first six chapters are to some extent introductory. Still, these may be the most important chapters in the book. These six chapters will change how you view application development for relational databases. In particular it will make you start looking at SQL a different way. You will look at a SQL statement, and start evaluating it almost immediately for its fitness to the EXADATA paradigm. Things like:

1) does the query contain constant tests that will allow EXADATA to use storage indexes to eliminate I/O?
2) will the joins in the query support pre-join bloom filter early elimination?
3) is the table partitioned (maybe by date) and if so are some partitions static such that EHCC (column compression) can be exploited?
4) does the query fetch data which when stored in a non-EXADATA environment, contains the inefficiencies that EXADATA SMARTSCAN can remove?
5) would the query benefit from less indexes (yep, you sill need indexes in EXADATA, particularly UNIQUE ones, but you need less of the others, maybe lots less)?

The first six chapters will introduce you to all the above and more.

The second best thing you get from the book is access to lots of new scripts for 11gR2 and for EXADATA. They give you a website you can visit. Cool Stuff and real timesavers. SNAPPER.SQL of course is the best. But my favorite is Name: ESFC_HIT_RATIO.SQL Description: Useless script for calculating an incorrect EXADATA Smart Flash Cache hit ratio (hehe), it seems they even include the junk for completeness.

No book is everything to everyone of course. One thing you won't find in this book is direct help on migration issues you will have with your applications and third party tools. Certainly the book gives information that can be used to work on problems you encounter. But this book is not a book of case studies. Chasing down answers to questions like why do my Informatica loads not all run faster? Why are many of my Business Objects reports which were timing out before, still timing out on EXADATA? There simply would not be enough room in one book to cover EXADATA directly, and these logical followon topics as well.

All in all, this is an excellent choice for an EXADATA practitioner. EXADATA is so new that there are so far only two books anyway. If you are going to be migrating, designing, or developing on EXADATA, you should get this book and if nothing else read and understand the first six chapters. Let it change the way you look at your SQL and your applications.

As an indication of the practical value of the book, consider this: I am currently lucky enough to be working intermitently with Tanel Poder, one of the books authors. This guy may very well be one of the Top-10 EXADATA practitioners on the planet. Every day we meet, he goes over various items, and much of what he talks about is the same material contained in this book. Other stuff can be derived from what is in the book. So this book is about real information and techniques, not just a recapitulation of the manuals. For example, Tanel uses daily, various scripts you will see in the book. SNAPPER.SQL is really good. He shows us how these scripts help us to understand what EXADATA is doing for us and possibly more importantly, what it is not.

UPDATE 29-NOV-2011

Here is an example of why this book works for people in the trenches. My boss recently asked me for proof that the EXADATA platform was benefiting the applications and database we were migrating to it. "Kevin, I need proof we are getting benefit from EXADATA before we continue to push systems to this platform.". Well... I knew exactly what to do. In demostrating EXADATA the book has pages that shows how to do this (for this question it was page 53 I think?). I went back to the chapter I remembered, read up to the page, and then ran the scripts and queries suggested. I then had the proof my boss needed. "See Boss, even on the poor side, our queries are getting a 5 to 1 lift in reduced I/O because of EXADATA. EXADATA is passing back only 1/5th the data it would have been required to move around on 9i.". He was surprised, and happy at the same time. So this is one example of what I mean when I say the book is for practitioners. The book gives you knowledge and then provides examples of typical ways you use that knowledge. It is the kind of book that changes the way you think and also gets you up and running.

Hail Flavius!

Written in a surprisingly readable style, this one took me off the fence regarding Exadata. Here you will find thorough research well-presented. If you are at all interested in this technology and would like to know more, and get it in a non-whitepaper, non-hyped way, this book won't disappoint. My congrats to the author on a great job.