Customer Reviews

Remoting Patterns: Foundations of Enterprise, Internet and Realtime Distributed Object Middleware (Wiley Software Patterns Series)

5 star:		(1)
4 star:		(0)
3 star:		(1)
2 star:		(0)
1 star:		(0)

 

 

Most patterns books are just plain junk; they're full of overhyped, overwritten trivialities. This one is not totally free from these flaws, but at least the topic is worthy and timely, and the writing is simpler and more to-the-point than in every single patterns book I've held in my hands (barring the original Gamma book). And I've seen them all. So, if you work with distributed systems, this book may be worth to thumb through, especially if you're a beginner.

On the other hand... there's a "but" here: first, I don't know if it's a translation, but the book is insufficiently edited; for a useage fanatic there'll be rich pickings here. Second, because of the first, you can understand this book only if you already know what it's about. Which can only be the case if you have already done a considerable amount of work in the distributed realm. Which, of course, may also mean that you don't need to read this book because you already know it all anyway -- although perhaps w/o the "patterns" slant.

So, I'm a bit ambivalent: the book is certainly better than the rest of the patterns literature, but who the target reader might be I'm at a loss to tell. I've worked with distributed systems since time immemorial and know all the platforms the authors talk about; and so, I didn't study it, I, rather, paged through -- and it did clarify/articulate a few things for me, but that's it; no conceptual breakthroughs, no design ideas. At the same time, a newcomer who could benefit will probably end up bogged down.

Another thing that I believe would be helpful for a newcomer and should have been included is samples of real implementations (meaning code, yes). Just looking at the pictures with arrows and surrounding disjointed text is not clear enough for someone who's never seen the thing itself. I've understood this book because I've read and written reams of such code and know exactly what the authors mean even when they express themselves in an a$$wise manner, but w/o such practical foundation the book will be hard to understand, I fear.

Anyway, it's an OK book. For what it is the cost is too high though, try to find a discounted copy.

PS. Btw, don't trust W.Boudville's review (below). If you want to know why, just check out his reviews page: the guy posts tons of reviews (nearly all five- or four-star ones) daily since the beginning of time: there's no chance in hell he's actually read all these books. And look at the inhuman breadth of his interests! C'mon, who are you kidding. There are many posters like that on Amazon these days, someone must be hiring them to post fake reviews. I don't trust any of those "Top 50 Reviewer"s anymore.

Patterns are one of the key concepts to emerge in computing in the 90s. But if you read a typical book on patterns, the issue of remoting patterns often gets scant coverage. In many ways, as this particular book by Voelter et al makes clear, the problems can be far harder than for a software package running on one machine.

At the most basic level, Remoting Errors (a pattern discussed early in the book) must always be a possibility as the communications channel between a client and a remote object can be subject to many failure modes. Then, later, there are issues of whether your object instances can be serialised into some kind of format able to be sent out over the network. You have to worry about marshalling and unmarshalling in the correct order. And the issues go on from there...

C programmers of client-server applications may be familiar with many of the topics, if they used Remote Procedure Calls. So too might C++, Java and C# programmers. The book's topics recur, largely independent of the programming language chosen, because the issues are fundamental.

The book gives a good coverage of .NET examples, and relatively little of J2EE. A bit more on the latter might have been good, perhaps.

CORBA gets a brief discussion. Not because it has many prospects of future growth, I imagine. But probably because it still has a substantial market presence. Due to all those poor blighters in the 90s who had to code in this miserable framework. If, hopefully, you do not need to use CORBA, you should still read this chapter. It's a salutary explanation of a brittle and inferior pathway, as compared to using XML and Web Services for a distributed application.