Reviews (11)

No Perfect Answer
From the start, I was happy to see that Barry wasn't one of those old school guys who tried something 10 years ago that didn't work, and had sworn off a similar endeavor forever as impossible. He's been there - he just has moved on, just as computing and software technologies have changed and made a lot of our past experiences less relevant today.

Neither is this book a recipe for perfection requiring strict adherence. His regular admonition to build for flexibility and his willingness to point out the flaws in conventional beliefs is a welcome departure from most architectural guides.

I particularly enjoyed the discussion on middle-tier architectures and data caching. As service-oriented architectures mature, it's highly unlikely they will succeed if all data access occurs in the enterprise tier. An SOA will encourage new applications and more sophisticated services, which will place even more stress on data access and responsiveness. Plus, the SOA infrastructure itself will require non-enterprise data for state management, diagnostics, object mapping and other functions that must be stored in the middle tier. This is a topic that cannot be avoided for an SOA that fulfills its long-term promise to the business. Barry's book is the only one I have found that deals with this reality in a straightforward, unemotional way.

Given that most of the SOA failures are likely to occur in the planning and organiation of the team or teams involved, I'm happy to see this kind of book now, when most companies are in the early stages. Let's let the standards firm up and the early implementations guide the "best practices" books which will no doubt come soon.

WebServices: dangerous material for inexperienced mind
The Beef
I wish there were 0 stars reviews. I wish there was a "R" or "NR" rating for books that would reflect appropriateness for certain "age" in one's career. This book ("Web Services and Service-Oriented Architecture" http://www.amazon.com/exec/obidos/tg/detail/-/1558609067) is dangerous in content as it is in its trumpeting up of the web services.

The Good Stuff
The only one good line out of the book is on p97: there is more money to be made on the services provided using web services than on the web services technology itself - yup you heard da-man: web services are NBYATP (nothing but yet another transport protocol) - pretty much since the dawn of geeks we witnessed healthy flow of trumpeted transport protocols that ended up at most as a 10 year fab - with TCP having the only staying power as the building block of our virtual lives. you know the the RPCs, the HTTPs, the DCOMs, the CORBAs, the IIOPs, the SOAPs, ...

Educator, Educate thyself
The author needs to get an education from "COM+ and battle for middle tier" (http://www.objectwatch.com/ and http://www.amazon.com/exec/obidos/tg/detail/-/B00007FYHG) where master Sessions explains why the whole shebang about caching data in the middle tier, coding transaction coordination by hand, and using stateful services is a KOD (kiss of death) for systems where number of simultaneous users exceeds one's age.

The author also needs to read more from the anti-middle-tier-christ Tom Kyte (http://asktom.oracle.com/ and "Expert One on One" http://www.amazon.com/exec/obidos/ASIN/1590592433) and understand why one shouldn't try to do what databases do best - process massive sets of data, quickly.

Delusions
The dangerous chapters start around p127 (sure, stick your data inside a middle tier access rules so nobody other then your application can ever get to it) and totally enter delusions on page 175 referencing magical "50 times or more" improvement using caching in the middle tier (sure that works for magical systems where no updates take place - find ONE real system where this occurs).

Distributed Caching
the distributed caching problem hasn't been solved at the academic level, let alone implemented in any of the OOTB systems out there. The less-then perfect approaches have been implemented (e.g. Network File System, Oracle Real Application Clusters, Google File System) but with their own quirks and for their own domain. These implementations just prove that distributed caching product is an NP complete problem and will probably stay at least until the next generation.

Transaction Coordination
for those who think that web services architectures will solve the multi-source data update transaction coordination, boy are you in for a ride. Sure, you can pick up one of the standards that Roger Sessions artfully keeps spitting at in his monthly newsletter where he displays the Cassandra syndrome by telling the "big guys" that you can't cache the data in any non-authority tier (usually non-database tier), coordinate transaction with stateful services *and* provide large volume support at the same time. The experienced will tell you that the more data you cache in the middle tier, the less clients you can service. Stateful web services - eeeek - that's the screeching of my fingernails against the metal wall - find me one business that has one dedicated phonecall taking service rep per client - so why do you think that anyone can afford to have a single instance of web service for one client only?

Stateless is the king
the obvious answer is that stateless services are the king, but they bring unfriendly problems that you know and love from HttpSession days - one call one transaction, can't daisy-chain them unless there is a transaction processing monitor environment available, can't rollback after the call completes, etc etc etc.

Caveat - Open/XA
One thing I will concede - if you have Open/XA compliant data source under the hood exposed via WebServices spec that supports recoverable Open/XA-like stateless transaction protocol (business workflow coordination spec) - yes you can fix all that's wrong with web services - if you can afford decresed throughput (in any measurable sense) AND inability to perform set operations.

Row-by-row security
just try applying security to the underlying data source. did you ever try to implement a security based on iterative approach where each "row" of information is fetched, and security access check is perfomed? then you know this works for systems where you fetch a few rows. try doing a search across say 10-15 million rows and apply your row-by-row security policy and you'll see why I wouldn't leave home without Oracle and its Virtual Private Database technology in my pocket. I can swifth through 250,000 users, 1M+ discretionary access controls (DACs, a.k.a. ACL) and 1M+ mandatory access controls (MACs, or functional security), and 10M+ documents in under one second - and all of it transparently available to middle tier applications, Business Intelligence (BI) reporting systems, and command line interpreters. what are YOUR numbers Mr. row-by-row middle tier WebServices security?

Maybe, just maybe...
web services will always be relegated to systems where flexibility and data transformation capability trumps volume in terms of simultaneous users, transactions, and data volume (but then you can use XSLT, perl, awk, etc); in other words once the number of simultaneous users exceeds your age combined with transactions volume above number of your fingers per second, and you can say good night to your web services.

Anecdotal
as a good friend of mine once said: "These guys built an XML over TCP transaction processing platform. They proudly stated they can process 3 transactions per second. Our production system at the exchange written in C blasts through about 10,000 of the same transactions per second. who are they kidding?"

Closing Words of wisdom
and to paraphrase Tom Kyte and add to this conversation: "I can do about 3,000 logged (recoverable) transactions per second from 5,000 clients on my ***laptop*** using a Java-JDBC-Oracle 10g all-in-one combination and about 15,000 nonlogged (nonrecoverable)transactions per second using PL/SQL as a client. What kind of volume can you service your web service again?"

5 Stars from the author of _XML:A Manager's Guide_
As a fellow author of a book targeted at managers, I feel that I have a pretty good understanding of their needs.Doug has done an admirable job of meeting these needs for Web services.This is a book for managers that want to have sucessful Web services projects.

Doug starts by motivating the need for Web services with a utopian view of a near future business trip.Then he gives a thorough account of Web services technology basics at a level that even managers whose technical days are long in the past can understand.He ties this account back to the utopian business trip, showing how Web services overcome the technical obstacles to making it a reality.

Reviews (11)

Don't purchase if you intend to migrate to 10g
This book contains a lot of material regarding BC4J and the jDeveloper IDE. Unfortunately, BC4J has been deprecated in 10g. [...]

Missed Opportunities
I received this book as a gift after I made it known I wanted to learn how to use Oracle's JDeveloper.When I began reading it, I winced over the cautions about who should read the book.Although I am proficient in Access and have done some VBScript and JavaScript client-side applications, I am new to J2EE and to Oracle.However, I hoped that reading the book would give me a good explanation of the basics of JDeveloper and that I could fill in the holes later.Unfortunately, the book did not provide many explanations of the basics and some of those attempts were pretty vague.For instance, it was not until somewhere shortly after page 400 that the distinction between Views and Entities became clear.In reviewing the earlier 'explanations' with the benefit of hindsight, they still did not seem to be very clear about what was just pointers and what held actual data.
The authors could, with very little additional effort, have provided short, clear explanations for basic concepts.That would have greatly improved the readability for a wider range of readers.It was this failure to provide better explanations of the basics that led me to title this review, "Missed Opportuniites".

Reviews (4)

Good intro, but not enough to teach you Oracle
I bought this book to get some info on Oracle 9i to see how it was structured. It was good for explaining the terminology and showed some screenshots. But this book will not get you started using Oracle9i, it only
helps you to understandsome of the syntax.

If you have a strong knowledge of SQL, then find another book, this book covers alot of java and xml explaination when dealing with 9i.

In short, if you are curious about using Oracle 9i, this book delivers
but if you want to learn the software, get another book.

hard to believe that this got published
Without wasting a lot of space on screen shots, this book provides step by step instructions for downloading and installing Oracle on both Linux and Windows and then using it.It explains new Oracle features such as the timestamp datatype and generally gets the novice up and running quickly.Everything else that you need is in the online/CD-ROM docs.If you're going to administer a production database yourself you'd probably want the Oracle Press dba book.If you're going to continue using Oracle you probably also want the bone-crushing Oracle Press "complete reference" book.But start with this one.

Reviews (61)

Disappointed
I am involved in a project to determine the best practices for integrating web services into our applications.We purchased a handful of recommended books on the topic including this one.I have to say that I am disappointed in this book.

First off there is little coverage of web services from an architecture point of view.There are some decent chapters that introduce the concepts of web services and SOA and then the authors jump directly into reference mode on the dozen or so technologies that they think you must master to develop web services.

Maybe it is just a failure of the J2EE process, but this book does little to address the confusion and complexity involved with developing web services in Java.In fact this book just adds to the problem.Developers should not need to know all the details of SOAP, WSDL, UDDI, and all the JAX APIs in order to develop web services.For the most part, all this plumbing technology should be hidden from developers yet it is the focus of this book.

Book needs a revision
This book disappointed me with belated material and using older java implementations. The content and examples needs lot of updates and confusing the readers. The book suggested website www.webservicesarchitecture.com is not working at all. I find no response from the publisher and author as well.

Editorial Review

If you've adopted Java as your organizational language, you're probably using, or planning to use, some sort of multitier design to maximize maintainability while making your data store accessible to as many applications as possible. The Jakarta engine ranks as the interface server of choice in that environment, and the Jakarta Struts Framework 1.1 makes it far easier to implement multitier information systems. Programming Jakarta Struts is the best how-to documentation around--in print or on the Internet--on the subject of using Struts to their greatest potential. Chuck Cavaness's book is comprehensive, detailed, critical of its subject where appropriate, and generally invaluable to anyone implementing the Model-View-Controller (MVC) design pattern in Java with the assistance of Struts.

Thankfully, Cavaness opens with an overview of the MVC pattern with a focus on how you're meant to implement it under Struts. For anyone thinking that implementing MVC sounds like more trouble than it's worth, this clarifies why such design usually pays off in the long run. After that, it's into the particulars, which include code listings (lots of them, delightfully commented) and crystal-clear block diagrams that show the flow of messages among objects. There are also many database schema charts that show how the authors structure data in the storefront and shopping cart application that spans the whole of this volume. --David Wall

Editorial Review

Patterns are higher-order designs that can be reused across projects and types of computer systems. Analysis Patterns: Reusable Object Models defines over 70 patterns, beginning with some from the business world, such as the Party and Accountability patterns, which define the players in organizations and whom they report to. Many of the other patterns are drawn from the health care industry and mainly show patterns of doctor-patient interactions.

The patterns for financial markets will probably be accessible for the majority of readers. Author Martin Fowler defines a Transaction pattern (and related patterns) as well as several patterns for the Accounting of Objects. He moves on to modeling stock markets with Portfolio, Quote, and Scenario patterns, which define how a price for a stock is determined for a given moment. Interestingly, he establishes patterns for Forward Contracts (for derivatives) as well as Options, and so takes on a complicated area in today's financial markets.