Why loosely coupled?

If you ever wondered why loose coupling is such an important concept in web services, this InfoWorld article by Jon Udell covers the ground comprehensively. Jon not only explains why developers need to couple loosely. He also sets out why many of them won't, and describes the extent to which tools such as BEA's WebLogic Workshop and Microsoft's Visual Studio .Net will help or hinder them.

This is a valuable article for anyone who wants to understand the dynamics of web services development choices. I particularly liked the concluding paragraphs, which deal with ideas about SOAP routers, application-level internetworking, and the need to hammer out trust mechanisms before the larger vision will work. Jon quotes a light-hearted yet insightful comment from Rohit Khare about where in the current 7-layer OSI model of the network stack these challenges may play out: "Khare joked that the real integration challenge lies at 'Layers 8 and 9 of the OSI stack: economic and political'."
posted by Phil Wainewright 3:55 PM (GMT) | comments | link

Why not aggregate collectively?

RSS newsfeeds make it possible to read multiple news sources at once, and now that many weblogs publish an RSS feed of their contents, you can aggregate the informal observations of your favorite bloggers alongside more formal newsfeeds into a single view. As a result, RSS aggregation is really starting to catch on.

There are a number of desktop products available that make this very easy to set up on your desktop — as described very comprehensively last month in this review by Jon Udell of Byte magazine. They can even subscribe to an XML feed in a single click, if the page you're viewing supports a newly popularized variant of the HTML link command. One of the most popular RSS readers is Radio Userland, which combines newsfeed aggregation with a desktop blogging tool, as Userland founder and long-term blogger Dave Winer describes in this article published yesterday on the Userland website.

What I've never understood is why everyone sees RSS aggregation as a solitary activity, never to be performed in public. Dave Winer quite rightly explains in his article that centralized aggregation doesn't scale, but Userland has gone all the way to the opposite extreme with Radio. When Jon Udell posts a list of the RSS feeds he subscribes to down the side of his weblog, it doesn't leave me any the wiser as to what they contain. What I really want to do is to skim their contents, exactly like he does, and I want to do it there and then, on his web page. Why doesn't he publish his feeds so I can do that?

In fact, I'd like to go one better than that, and be able to read an RSS feed of his RSS aggregation, because then I could get an at-a-glance view of what he's viewing without even having to go to his blog in the first place (unless I saw a headline that made me want to read the whole story, in which case of course I'd click through to his site). Maybe if this idea caught on, he could slim down his own list of RSS feeds, because surely someone somewhere would start aggregating some of the feeds that he subscribes to, and then he'd be able to subscribe to just one feed in place of half-a-dozen. And you know, with that step, we've just started to venture into the realms of network leverage, because the expert he's relying on to aggregate those feeds for him would automatically add new feeds as they became available, so Jon would be adding valuable new feeds without even having to lift a finger.

It was always my intention to create an aggregated feed like that as part of this website, and a few days ago I finally completed the first iteration. The Loosely Coupled news page, which launched last weekend, publishes the most recent feeds from sites that I follow for information and comment about on-demand web services. Rather than offering an aggregated feed of the entire page, which requires further software development, I'm demonstrating the concept of a feed-of-a-feed by picking what I regard as the most notable stories or comments and publishing them both as a feed and also as a Javascript insert, which you can see both on the news page and on this page too, in the gray right-hand sidebar.

I'm surprised the folk at Userland haven't explored this way of doing things, because it seems so much more in tune with the spirit of blogging than the current model of private RSS aggregation. I add another layer of surprise when I consider that there may be astonishing synergies to be gained from managing tiers of aggregation using OPML, an XML-based outlining format that Dave Winer created and is fond of for organising hierarchical directories. Maybe all the pieces will fall in place soon.

You may well ask, why don't I use services like NewsIsFree or Moreover? There are two reasons. First and foremost, I agree with Dave Winer, they don't offer the flexibility of being able to directly subscribe to whoever I please. Second — and this is quite important — I'm not allowed to. Loosely Coupled is a commercial website (although, as my wife is constantly reminding me, the revenue-generating elements are yet to make an appearance, but that's another story). As a business, I'm happy with that, because if I'm relying on critical services, I prefer to be paying for them. But becoming a paying customer means going to the back of the queue and waiting for their salesperson to call and sell you a contract, which usually takes a very long time if they haven't heard of your company. I contacted both of them a week ago. I've heard nothing yet.

While NewsIsFree and Moreover are busy chasing the corporate market, I have no doubt that someone will come in and steal the real opportunity away from them. Maybe that will be the new venture that Moreover founder Nick Denton is currently developing in stealth mode. More likely, it will spring from initiatives like UKOLN's RSS-xpress Lite, or maybe even from individuals learning to use PHP and doing it themselves, as I did (gratefully aided by the online assistance and starting-block code of Julian Bond and Kevin Yank).

I'll be documenting the code I've used for the Loosely Coupled aggregator in my Appswitching Diary blog over the next few days for those who are curious, but I should add that the architecture is not so much loosely coupled as Heath Robinson at present, so don't hold your breath. The purpose was not to write software, but merely to demonstrate what I believe will become a very powerful application concept.
posted by Phil Wainewright 7:58 AM (GMT) | comments | link

Giving the Web a human face

"Where are the business advantages in becoming a 'faceless third party'? Even if you're the most reliable faceless third party in your particular application niche?" asks Leigh Dodds while letting off steam about UDDI.

Because the Web runs on computers, we sometimes forget that the most important nodes in the network are the people who sit in front of the computers, tapping away on their keyboards. They make business decisions based on all kinds of parameters, many of them as yet unmapped by science, and few of them fully encapsulated in software. As Esther Dyson has noted in a much-quoted article in the latest edition of Fortune, "When you put up a website, you're not erecting a billboard, you're opening a door, and people come crowding in. You have to have the staff there ready to greet them." The only web-based business processes that are worth thinking about are those that enhance and enrich those human interactions.

Back to Leigh for the last word: "While I agree that a web service framework that enables more automated business processes across the web is desirable, I don't think we're ever going to be in the situation where you're going to be working with faceless third parties. It's not in their interests, and it's probably not in yours."
posted by Phil Wainewright 6:35 AM (GMT) | comments | link

Understanding orchestration

Today, we have applications. In the future, all we will have is orchestration. Today's applications will have atomized into discrete, distributed components within an all-encompassing web services architecture. Orchestration provides a means of combining those disparate components into joined-up process flows that will automate end-to-end business functions. It's where the business biosphere meets the global computing grid.

Explaining orchestration is a complex task, but an extended article by Stuart Johnston in XML & Web Services Magazine does an excellent job of introducing the important concepts, along with some of the emerging industry players. There's also an interview with David Kiker, Microsoft's general manager for e-business servers, about how orchestration fits into Microsoft's vision. Kudos to Collaxa, a leading orchestration startup and also one of the few vendors in web services-land to maintain a weblog, for the link to these articles. Here are three definitions of orchestration that I've extracted from the two pieces:

• "In the new model of the agile enterprise, software should be componentized for easy reuse and adaptation in service-oriented architectures. ... Orchestration technologies make it possible to logically chain discrete functions into interenterprise business processes" — Stuart Johnston, author of the article


• "The developers that have to build and tie all those services together are exposed to a set of new challenges related to sending and coordinating conversations with those remote services. It is the handling of those loosely coupled conversations that we call orchestration." — Edwin Khodabakchian, CEO of Collaxa


• "The role of process orchestration has been undervalued by the computer industry as a whole ... I believe that business process is just as important as business data, and that at some point in the future there are going to be applications on people's desktops that are as common as Excel that allow people to deeply understand business process in the same way that they can deeply understand the data." — David Kiker, Microsoft

posted by Phil Wainewright 4:45 AM (GMT) | comments | link

Putting UDDI into context

People have wrongly perceived UDDI as the Google of web services discovery, writes Doug Kaye. A closer analogy might be Bigfoot, founded in 1995 as a global 'white pages' directory of Internet email addresses. Except that UDDI always was intended to serve as a foundation standard, on top of which competing providers and vendors can build more sophisticated services. "The problem is that we've placed UDDI at the wrong point in the protocol stack," writes Doug. "It's not the equivalent of a Google; it's the equivalent of the Web ... As with the Web, there's another layer of value-added services that will organize, rank and otherwise make sense of this data."

Doug highlights SalCentral, one of the first commercial web services brokers, as an example of such services, and goes on to note that "SalCentral has been testing a new in-house tool that crawls UDDI the same way Google crawls the Web" — which sounds very interesting.

SalCentral (which in case anyone is misled by a typo in Doug's posting I should point out is based in the country of Wales, and not in a place named after a species of ocean-going mammal) is taking advantage of another of those business opportunities around UDDI that I mentioned last week. Here's another useful backgrounder about UDDI I came across since that posting, part of a set of articles from Network World. A companion news story describes how Novell, Sun and CA are working to create UDDI registries based on their respective directory services products, fusing UDDI with the LDAP standard for Internet directory services.
posted by Phil Wainewright 3:38 AM (GMT) | comments | link

So tell me again, how do we make money from this?

At last, someone has noticed the missing ingredient in the web services stack: billing. Everyone has been talking about subscribing to web services without ever giving a single thought to one of the most important things that happens when someone takes out a subscription — they hand over a fee. "Without billing, it's just a hobby," Scott Swartz, CEO of web services billing software vendor MetraTech, reminds us in this story from InfoWorld. "Companies need to focus on collecting revenue to make a profit from Web services." All credit to Cape Clear for noticing what everyone else seems to have forgotten, and sealing a product deal with MetraTech, which seems to be about the only company out there with proven technology that plugs this all-important black hole. Billing's not easy. No-one can afford to make it an afterthought.
posted by Phil Wainewright 2:49 AM (GMT) | comments | link

Novell brings cheer with SilverStream buy

The news that Novell is to buy web services platform vendor SilverStream Software settles several longstanding questions.

The most important of those questions concerns Novell's strategic direction. In a recent posting, I cast doubt on whether the company's management had its finger on the pulse of the web services sector. I was wrong. This acquisition shows that Novell's top people are very much on the ball. Of all the application server vendors, SilverStream is the one with the most coherent family of web services tools, and it handsomely plugs the gap in Novell's current lineup.

Once the deal goes through (it is expected to close in July), it could be the catalyst to catapult Novell into the front rank of leading platform vendors. There may be more to come, too. Novell has spent $212 million on SilverStream, but its rumored war chest was as much as $700 million, which leaves a cool half a billion potentially still in play. That would be enough to pick up HP's middleware properties, which HP is thinking of selling, or perhaps even make a bid for market leader BEA. But it now seems that Novell is smarter than that. Although there may be grounds for acquisitions to gain market share, its best strategy now will be to conserve its cash to pick up some of the startups who are developing much-needed new web services tools and technologies.

That brings us neatly to the second of those three questions. This one has been hanging in the minds of investors who've been putting their money into web services ventures. Did they do the right thing? Novell has answered in the affirmative, demonstrating that there is real value in solid web services businesses, and inspiring confidence that there are substantial cash buyers ready and waiting to acquire a market presence.

The final question concerned SilverStream's future. Along with Iona, whose history I wrote about last week on ASPnews, SilverStream is one of the last survivors of the several dozen web application server startups that came to prominence in the mid 1990s. Almost all its peers were acquired, the most successful of them by Novell's soon-to-be rivals BEA, Sun and Oracle (IBM and Microsoft also made acquisitions, but not of core platforms). It seemed destiny had passed SilverStream by, but then the advent of web services allowed it to display its strengths in a new light, and its persistence has finally been rewarded.
posted by Phil Wainewright 11:33 AM (GMT) | comments | link