Customer Reviews

Oracle Tuning: The Definitive Reference (Oracle in-Focus Series)

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

 

 

It would be easy to criticize any book on such a big topic for being incomplete. The problem here is that this book leaves out essentials that the authors have to know and just didn't bother writing down. Here's an example. "The parameter named optimizer_index_cost_adj controls the CBO's propensity to favor index scans over full table scans." Well, that's great and it's something I needed to know. So what are the values and what do they mean? You'll have to go look that up on the web. This happens over and over and suggests to me that no one bothered to read the book critically.

Now I've had the book for a few days and it keeps getting worse - if I were writing my review now it would be at 3 stars and heading down toward 2. There are SO many mistakes. There are sentences with extra words in the middle - no one ever read this stuff. There are scripts that just don't work, that have SQL errors. There are pompous bits that don't say much strewn here and there. In spite of the book's apparent top-level organization, the chapters themselves are pieced together carelessly with very little development of any ideas. Bits of a given concept are referred to here and there with no attempt to tie them together. One chapter cites one value for an init parameter, another mentions a second value, but nowhere are all the values explained or compared. There's still a lot of information but some of it is hardly usable because it's incomplete, or wrong. Maybe one star....

To add insult, each script contains a warning that it can't be used for a commercial purpose without licensing it from these folks! Who on earth do they imagine is buying these books? Schoolkids? If the scripts worked this would be really sad, but as they don't it's not much of a loss.

I like to think that if I get one really good idea or strategy from a book it will pay for itself, but "Oracle Tuning: The Definitive Reference" is causing me to re-think that philosophy, because it's such hard work wading through all the mess.

In contrast, I'm reading Jonathan Lewis's book on the cost-based optimizer. It's lucid, coherent, well developed, and very usable. Even his old book on 8i is far more useful to me than Burleson!

I had high hopes for this book but as I worked my way through it I kept finding myself attempting to puzzle out the author's meaning because of typographical errors, poor proofreading and poor editing. Sometimes I literally could not see the connection between what the text was saying the sample output was and the sample output that was printed on the page. I expect a technical book to be technically clear and this one let me down too many times to instill confidence. I also came away with the sinking feeling that it was at heart a grand advertisement for the WISE software; another confidence killer. This is my second Rampant book and I think my last.
To be fair, I learned some new things and will undoubtedly improve my skills because of this book, but I would say to the authors and publisher that you made me more mad than you made me smart. You have lost me.

Overall, I found this book well worth the read. As with most rampant books I've read, it has the following consistent issues:

Lack of organization and proofreading
Too bloated (redundant information, large fonts and too many pictures)

Additionally, I found numerous errors in the code depot scripts that required reworking for them to function properly. That being said, there are some really great nuggets of wisdom here as well as some key concepts that DBAs should be aware of. The methodologies are usually solid and the information is often, but not always correct.

One thing missing from this book that I find in the works of Tom Kyte and Lewis are the actual examples. There are so many good ideas here, but no before and after statistics to give the DBA a feel for the validity of the arguments. Strong test scenario scripts provide proof to the theory and give the learning DBA a chance to practice through example, not just study the concepts. And this is woefully lacking in this book.

Overall, I'm glad I read it and I did gain some insights, but it could have been much better. Take these theories and then dig into APress works for actual applications and some revisions. And take everything with a grain of salt until you prove it yourself.

This is a well written and good book targeted for senior DBAs. I gained a very good understanding of the Automatic Workload Repository (AWR) reading this book. But there is a huge and fundamental flaw in many of the scripts used throughout the book. Let me explain this:
Most the tables in the AWR stores cumulative data for various Oracle statistics. If you think about it, it makes sense because Oracle can get the delta between two snapshots by doing subtraction. The scripts in this book does not consider this fact. I fail to understand how the authors can miss this point. Because of this the output of most of the scripts is wrong or meaningless. Of course, it is not a big deal to fix these scripts ourselves. But the fact that the authors missed this fundamental point annoys me and that is the reason for single star rating from me. I will be happy to be corrected if the authors (or some other oracle experts) can explain that scripts are indeed correct.

The book has a wealth of ideas on tuning and queries for making sense of Oracle's tuning related views.

The book suffers a bit in all of the following areas: a lack of organization, poor editing, a few blatantly incorrect statements and not enough detail.

One of the owners pushes his graphical tuning software. I'm not big on that - as other much more mainstream tools are ignored.

While reading the book, I got the feeling it was a hodge podge of thoughts thrown together as fast as it could be written. The authors obviously split the chapters and then it looks like they didn't read each others chapters. Some material was repeated and even that was still superficial in detail.

I haven't run many of the scripts but, just throwing the ideas and concepts in to one place for reference along with the scripts is probably worth the price of the book.

All in all, my opinion, this is not one of Don's best works. He obviously didn't spend the same amount of time on this as he has on past works.

In the books defense, putting the detail in that I point out that is missing, would have added another 1500 pages to the 960 that are published.

The book is a 3 star piece of work, but I'm NOT sending it back either.

I own 5 or 6 Rampant books. I find across the board that they have nearly-useless indexes. Specifically, it is difficult to tell if the subject is covered at any depth, as there are no ranges (e.g., 445-450). Instead the index will say 445, 446, 450. I end up using the table of contents and wasting a lot of time.

Before I get slammed by comments ... This might seem like trivial matter, but compare the indexes in the O'Reilly books, which are much more useful.

(I used to write indexes professionally).

As someone who is working hard to learn the ins and outs of Oracle, I've acquired a fairly large number of books. It doesn't bother me to note that the authors stumble a little with the language and grammar as it's a logical conclusion that English, in today's world, isn't everyone's first language. I know the language; I want to learn how to make my Oracle database as efficient as possible. This book holds a tremendous amount of information and is well worth the cover price. I've tried the scripts, but I never took it to mean that "script" was short for "scripture" and I didn't buy a Tuning Bible looking for answers without having to do the work. I bought a Tuning Reference, and as such it's never more than an arm's length away.

Very good, enough internal tips, common sense tips etc. I have read those parts that are interesting for me like AWR/ASH and some tuning spots. I got some good ideas. I have done work over 15 years with Oracle and some other databases. For me this is proper book, specially AWR and ASH. Perhaps even more internal tips which are gained over years pro work would have been waited. Okey, consult must have something of his own, when visiting customer;)

This book is a large reference volume, of about 1,000 pages. Rather than trying to read this large book chapter by chapter, I recommend skimming over the sections at first, to see which areas suit your particular needs. For instance, you will likely find Chapter 16 of great value. It's entitled "Oracle 10g Wait Event Tuning with AWR and ASH." Or, if you are working with data warehouses, you might check out Chapter 18, "Oracle Data Warehouse Tuning."

Although there is much discussion about tuning in general, I think its great strength is the focus on Oracle 10g Tuning--especially "time series metrics," using the Oracle 10g AWR (Automatic Workload Repository.) In fact, I found the most valuable chapter to be Chapter 9, "The AWR Time Model Approach." In this approach, the DBA determines where the database is actually spending its time. After all, that's what performance tuning is all about--where's the time going. The authors explain how to use the time performance views: V$Sys_Time_Model and V$Sess_Time_Model. These views gather cumulative stats for either the entire instance, or per session.

There are also references to more traditional tuning techniques, such as gathering proper statistics, SQL tuning, join order, etc. The authors also include chapters on Network Tuning, Disk Tuning, as well as tuning the various cache buffers and pools. Chapter 19 explains how to do performance tuning with the aid of OEM (Oracle Enterprise Manager.)

Of special interest to OPS users: Chapter 20 provides an overview of RAC (aka Oracle Parallel Server, OPS). The authors illustrate the "Cache Fusion" architecture, and discuss the "data localization" load balancing method.

This brief review doesn't come close to doing justice to the huge amount of valuable information in this book. The discussion of the DBA_HIST views is alone worth the price. This is likely the most complete set of Oracle tuning information in a single book. You really have to see it to believe it!

P.S. I neglected to mention the excellent sketches by Mike Reed. A very nice touch!

Despite new "self tuning" features in recent versions of Oracle, database tuning continues to be an essential part of the DBA skill set, but where do we acquire these skills? There is no substitute for experience, but once in a while there's a roadmap for it.

In their new book Oracle Tuning: The Definitive Reference, Alexey B. Danchenkov and Donald K. Burleson reveal a holistic, platform agnostic approach to tuning the Oracle RDBMS. Both proactive and reactive tuning are given ample treatment while always conveying the "why" and not just the "how". The techniques presented are complimented by a free copy of the Workload Interface Statistics Engine (WISE) tool (available via download), written by Danchenkov, which provides an interface into the tuning tables and views in Oracle.

The authors, clearly tempered by years of experience, take a very realistic approach to database tuning. They acknowledge that the DBA may not have the time, ability or influence to bring upon an application rewrite or change in server architecture. The bulk of the book focuses on tuning methods within the realm of the database administrator (though all areas affecting Oracle performance are covered.) While focusing on Oracle Database 10g the authors present tuning concepts and techniques in a way that many of the techniques and nearly all the concepts are applicable to all Oracle RDBMS versions.

Thoroughly covering everything from disk to SQL the book is littered with the exact commands you will be running in the field including example output and analysis. The authors have also included several pages of "Silver Bullet" tuning examples. These examples demonstrate how a quick diagnosis and the right tweak can save the day.

Testing a hypothesis on a large active database is like trying to tune a car while it's flying down the freeway at 75 miles per hour.

This book is not for the beginner. If you do not feel confident about your knowledge of the Oracle architecture you will feel overwhelmed by this book. Of course if you do not feel confident about your knowledge of the Oracle architecture you should not be tuning a database.

For those comfortable with Oracle but new to tuning there will be many paragraphs you will read, re-read, then read again, but Danchenkov and Burleson have not missed a step. On almost every topic there are a couple notes on common pitfalls and how to avoid them. The authors have really taken great care to shepherd you safely through all steps of tuning the database.

In barely less than 1,000 pages, Danchenkov and Burleson have compiled the definitive reference for Oracle tuning. Coupled with a good background in Oracle, this book contains everything you need to tune almost every aspect of the Oracle database. I highly recommend it to the Oracle professional looking to learn about tuning or the experienced tuner looking for a good reference. The type of tuning presented in this book could easily lower your hardware costs and make you a rock-star DBA.