SOA in a Nutshell

I'm still struggling to get my head around Service Oriented Architecture and all that it implies. So far, it has only been very peripheral to what I do, so I'm ever on the outlook for pearls of wisdom about SOA.

Neil Ward-Dutton of Macehiter-Ward-Dutton has come up with a very nice visual summary diagram of the key benefits of an SOA approach at SOA's five benefits in one picture (click on the image there to see a legible enlargement, below is merely a thumbnail).

Web 2.0 -- on a wing and a prayer!

"On a wing and a prayer" ...

• In poor condition, but just managing to get the job done. (from The Phrase Finder)
• In a desperate situation and you’re relying on hope to see you through. (from World Wide Words)

- - - - - - - - - - - - - - - - - - - - - - - - -

I've written a few posts about "Web 2.0" during the last year or two, and consider that I have a reasonable understanding of the concept -- even if I reckon, a little dismissively perhaps, that it's sort of like the notion "This is the place to be!" and will inevitably be replaced by some other faddish concept.

Another current favorite term is AJAX (Asynchronous JavaScript And XML): not an architecture or a product specification but an approach or very general methodology for designing and developing Web applications. Some of them, like Google Maps, really are very nice! But already there are dozens of incompatible developer toolkits (I've listed a small sampling of them at http://www.asiapac.com.au/Links/WebServices.htm and http://www.asiapac.com.au/Links/Design.htm#AJAX_Asynchronous_JavaScript_And_XML.

Then there's SOA (Service Oriented Architecture) not to forget Web Services, with an "acronym soup" of terms (such as SOAP, WSDL, UDDI, ...) -- indeed so many and varied are they that the pithy catch-all WS-* is generally used for convenience. Some people treat the terms SOA and Web Services synonymously, so there's a degree confusion or misinformation going the rounds. (They're not synonymous, but Web Services can form part of an SOA design.)

Like others, I'm getting annoyed by all this excessive hype about "Web 2.0" and I've already predicted its inevitable demise ... See Web 3.0 is Underway -- but "Web Pi" is unreachable ... I like the graphic so much that I'm repeating it here:

Click here to view the Wikipedia article about Pi

Apparently somebody else quite independently has come up with the same construct. See The we is evolving into web pi (though without having fleshed out the concept very far, only proffering "pi is good as u can C").

All that aside, it was two web interactions earlier today that irritated me top the point that I had to vent some steam in this blog.
To me, the most ridiculous thing about "Web 1.0" (not even Web 2.0) is that it's based on the fragile Web Browser as the delivery vehicle, rather than some sort of session-aware rich client.

And, as often happens, it was only a little thing that drove me over the edge today: I was carrying out two extremely simple browser-based transactions.

Firstly, while doing some Internet banking at the ANZ Bank website, I quite inadvertently pressed the Back button on the left side of my mouse (the result would have been the same if I had pressed the Back button near the top of the browser window). You already can't say what happened, can't you, because it has happened to you too? I got an error message saying, in effect, "You idiot, you are not allowed to use the browser Back button at this time. For your safety, you have been disconnected. Please log in again and restart your transaction. (By the way, did we tell you that you're a moron?"

How repugnant this is! How can they keep getting away with this when another Australian banking institution's site that I also use for Internet banking gracefully takes me back as page, as I expected. Is it not fair to expect that if one bank can do it properly, they all should? (By the way, apart from this fundamental design flaw I find the ANZ Bank's site quite nice to use.)

Secondly, a little later in the day, when three or four pages into filling out a multi-page questionnaire at the Australian government's Centrelink site, I inadvertently did the same thing again. Fancy me being so naive to expect the Back button to take me back one step here, either! The principle is the same, however in this case I was told:

Error 500: Unable to restore flow execution with key '_coed-2907-2533-8831-4187A77DB2BA_k0512D7BE-125B-2129-A2FA-5C769C52F9EB' -- permission denied.; nested exception is org.springframework.webflow.execution.repository.support.InvalidContinuationIdException: The continuation id '0512D7BE-125B-2129-A2FA-5C769C52F9EB' is invalid. Access to flow execution denied.

So until they sort out some simple architectural matters such as proper handling of the browser's Back button, why are they progressing at breakneck speed toward "Web 3.0" or whatever comes next at the bleeding edge of the hype cycle?

The whole issue of browsers (in all their flavors) versus "fat clients" or "thin clients" with their various advantages/disadvantages needs to be more fully worked on. And even if the industry moves toward SaaS (Software as a Service) with centralized Web processing via ultra-fast broadband, there's still no way that I would carry out certain types of activities on anything but a local Rich Client platform, with local control (of backup, security, performance, etc). Well, them's my preferences; I fully understand that yours might be different.

Understanding SOA Security

Earlier on int this blog, I've made several posts about Service Oriented Architecture, such as SOA means Business! or SOA's Business Value and also a couple of glib posts about "Web 2.0" such as Web 2.0 approaches.

(SOA is often mentioned by some commentators in the same breath as Web 2.0, but it's definitely not synonymous. In fact, SOA applications don't necessarily even have to run across the Web, though in many of not most cases they will. I'm obviously not the only one who thinks this, as indicated by tow recent posts: Can, or should SOA be implemented without web services? and TRUE or FALSE: SOA cannot be implemented without web services? )

Not to forget my own concept of "Web Pi" ... see "Web 2.0" and "Web Pi" -- Reject Reality and Substitute Your Own! By all means feel free to use the term and espouse the concept (if you're game), but it's a cynical enough concept for me to have a deep-seated psychological need to keep attribution to it!

Anyhow, one of the essential requirements of SOA is that all services must observe appropriate security.

And if you want to learn more about this, you'll be hard pressed tot find a better source than the new IBM Redbook Understanding SOA Security: Design and Implementation (Like all Redbooks, it's free.) So go download it and have a good read!

Which one is "bigger" -- Microsoft or IBM?

There's an IT industry debate that has been going on for a decade or more now, about whether or not "The mainframe is dead."

IBM certainly doesn't think so, and they are making a concerted n effort to get young IT people interested in mainframes and all they can offer. See, for example, these eWEEK reports: IBM to Spend $100 Million on Mainframes and Mainframers Learn New Tricks and Long Live the Mainframe.

It seems that part of Their campaign to rebuild interest in the field involves several new IBM Redbooks. The first of these is Introduction to the New Mainframe: Large-Scale Commercial Computing and here's part of the abstract:

Today, mainframe computers play a central role in the daily operations of most of the world’s largest corporations. While other forms of computing are used in business in various capacities, the mainframe occupies a prominent place in today’s e-business environment. In banking, finance, health care, insurance, utilities, government, and a multitude of other public and private enterprises,
the mainframe computer continues to provide the foundation of large-scale computing to modern business.


The reasons for mainframe use are many, but generally fall into one or more of the following categories: capacity, scalability, integrity and security, availability, access to large amounts of data, system management, and autonomic capabilities. This IBM Redbook is designed for readers who already possess a basic knowledge of mainframe computing, but need a clearer understanding of how these concepts relate to mainframe planning, implementation, and operation.

There are a couple of other associated Redbooks:

• Introduction to the New Mainframe: Networking
• Introduction to the New Mainframe: z/OS Basics

IBM has been providing IT products and services to the enterprise for many decades now (for at the very least three decades before I joined them in 1970). And with Service Oriented Architecture being all the rage these days in the enterprise architecture space, here's another IBM Redbook that's well worth reading: SOA Architecture Handbook for z/OS

- - - - -
Well now, it's February 2007 and statements are being made that Windows Vista is "the biggest operating system ever written". Using what metrics, I wonder. (If I were a scalawag, I would say "biggest amount of hype" -- but as a mild-mannered reporter I resile from saying that!)

The z/OS operating system has its roots in OS/360, going way back to 1964, so there's decades of accumulated wisdom and knowledge in the z/OS operating, and in the associated IBM mainframe hardware architecture. In my opinion, for enterprise-scale computing the cruddy 32-bit and 64-bit Intel style hardware architecture are quite inferior, as are the Windows operating systems. I'll agree that Windows is getting better all the time, but so is z/OS (still being tweaked some forty years after its conception).

Another product range that is dear to my heart also comes from IBM: currently it's known as the IBM System i (with its operating system being called i5/OS). Before that it was called the iSeries, and before that the AS/400 (with operating system OS/400). Its progenitor was the IBM System/38 (with operating system call CPF - Control Program Facility), conceived in the early 1970s and announced in 1978. But its essential architecture is still very much alive and well as the System i, with major enhancements being made in each new release.

One thing that really irks me about the Wintel platform is the hiatus between 32-bit and 64-bit modes. I would like to use 64-bit Windows XP on my brand-new dual core AMD desktop system, but I'm simply not prepared to since I would have to go through purgatory due to lack of device drivers, and on top of that doubtless have to buy new 64-bit versions of applications (if they existed, and in many or even most cases they don't). I clearly recall that when the System/38 and later the AS/400 changed their internals to go from 32-bit to 48-bit and later 64-bit mode, customer applications were automatically adjusted by the system to run in the new mode without any application redesign whatsoever and a minimum of fuss. Superb! In comparison, the Wintel approach is quite horrid.

Anyhow, getting back to my original theme: Which one is bigger, Microsoft or IBM? Microsoft certainly is bigger in terms of overall annual software revenue, yet I'd say that IBM is bigger in terms of the considerable range of software products that it offers. Microsoft has a strong research arm and generates lots of patents annually, but I think IBM well and truly beats it with some ground-breaking fundamental physical research (and more patents annually).

To take an area of software where IBM has always been strong, namely database. Even venerable hierarchical database IMS is still in production, but around 1980 IBM released its first commercial relational database DB@, and it has gone from strength to strength. Not to denigrate Microsoft;s current versions of SQL Server (which are very competent indeed), but IBM's latest DB2 9 release (code named "Viper") with deeply integrated native XML support are quite superb. See for example DB2 9 for z/OS Roars to Life and DB2 9 pureXML Guide

I notice that one favorite part of IBM has reached quite a milestone. See the Special report: Celebrating 50 years of the IBM Journals

Since the first publication of the IBM Journal of Research and Development in 1957 and the IBM Systems Journal in 1962, these Journals have provided descriptions and chronicles of many important advances in information technology and related topics ranging from atoms to business solutions. To celebrate the 50th anniversary of the IBM Journals, this report highlights a selection of significant papers published in the Journals, along with brief commentaries. The Journal editors chose papers which were very highly cited in the technical literature, described technologies of historic significance, or provided an important overview of a field.

This would make a scalawag (not me) ask what Microsoft was doing in 1957 ...

So, which one is "bigger" and how do you define bigness? Who cares, some of you will be thinking, but if you're interested why not have your say!