Adopting a componentized software architecture is one of the key success factors identified by IDC in its latest survey of the web-native application service provider market, which it says will grow from $200 million in the US in 2001 to over $1.5 billion by 2006. Personally, I was always rooting for the web-native ASPs, since I didn't see much future in trying to deliver conventional enterprise applications over the Internet. Unfortunately, most of the industry disagreed with me, with dire results. So it's good to see companies like NetLedger, Salesforce.com, WebEx and Upshot finally winning through. Many of them wisely disassociated themselves from the ASP moniker once it had been hijacked by the software industry establishment, but to me they were always the ones that properly understood what it really meant, while everyone else missed the point.
I've mentioned in previous postings how switched-on IDC has become about the implications of web services architectures for applications, and this report of their findings provides further evidence: "Web-native ASPs are particularly suited to creating, aggregating and delivering self-describing and discrete web services applications as components that can be called on demand," said IDC research manager Karen Moser. "As new applications emerge based on the Web Services Architecture (WSA), web-native ASPs should be at the forefront of this trend ... In a broader sense, everybody will eventually be web native."
It's not all praise for IDC, however. Another of the research group's predictions reported this week sees the market for professional services related to web services projects rising to $7.1 billion by 2006. Having been round the block with IDC and other analysts during the hype years of the ASP industry, I know only too well why we're seeing these huge numbers being quoted. IDC is a business just like any other, and right now there's a demand for nice, fat growth figures that entrepreneurs and business managers can put in their business plans. So IDC goes round and asks the vendors what market demand they're seeing, and they speak to customers to ask what they expect to do, and everyone has their rosy-tinted glasses on right now, so the answers that come back project stellar growth.
IDC is just doing its job when it puts those results into a bullish market projection. There's no malicious tampering with the figures here; it's simply a well-established process of mutual self-delusion. Yes, web services is the future and yes, it will be big business. But no-one ever said it was going to be easy. If it was totally predictable, it wouldn't be an opportunity. With opportunity comes risk, and you cannot mitigate those risks by putting your faith in research projections, not even if they come from IDC.
Keeping it to less than a dozen words is a good rule of thumb for an elevator pitch. The essentials are captured perfectly in the following sentence, I think:
"Web services are self-contained business functions that operate over the Internet."
Not a word about software or standards those are implied anyway, because it doesn't work without them and instead it focusses on the part that really matters: "self-contained business functions". Follow up those dozen words with an example that makes sense in the context of the listener's business, and then stop talking. If they're hooked, they'll start asking questions. If not, you're the person they'll come back to later on when it finally clicks with them.
The quote comes from a much longer article on WebServices.org about "conveying the essentials of Web services to your CEO". It's good stuff, and well worth rehearsing in preparation for those follow-up questions. But don't put your trust in delivering all of it at once, in what the writer refers to as an "elevator speech", because the boss will have gotten out long before you reach the end of the third sentence. The only speeches that get delivered in elevators are soliloquies. If you want an audience, limit your pitch to those perfectly succint first dozen words.
posted by Phil Wainewright 3:02 AM (GMT) | comments | link
Thursday, June 06, 2002
UDDI is not enough
A groundswell of dissatisfaction has been rising against UDDI, and not before time. A very thorough article in this week's edition of Infoconomy Collaborate gives a detailed overview of the current state of play. As I mentioned when noting the concept of service grids back in April, "UDDI ... merely lists web services the same way a yellow pages directory lists minicab firms" or, for the benefit of our American readers, perhaps I should say pizza delivery outlets. In other words, UDDI lists services, but with no certain guarantee of quality, reliability, timeliness or persistence. And no refunds either.
These shortcomings open up significant business opportunities for enterprising ventures to become specialist UDDI service providers probably leveraging directory services technology, as the Infoconomy article notes who will verify and guarantee services beyond the basic information provided under UDDI. One of the first startups on the block to exploit this opportunity is Infravio, of which I'm sure we'll be hearing more in the future. Infravio's platform addresses two specific failings, as its VP of product marketing Richard Petersen explained to Infoconomy: "UDDI cannot manage changes to one web service across to other related web services, and it can not manage semantic differences between web services." Covering those gaps is a start, but many others still need to be filled.
posted by Phil Wainewright 3:02 PM (GMT) | comments | link
Wednesday, June 05, 2002
Embracing and extending the desktop
Networked services add to, rather than eliminating, client functionality. This is in contrast to the early beliefs of network computing purists, who, in the name of manageability, sought to remove all functionality from the network edge. First-generation ASPs took the purists' message to heart, and consequently failed to gain the confidence and trust of users, who didn't want to compromise the rich desktop functionality they were used to.
The highly distributed component architecture of web services allows server-based services to complement and enhance client functionality, at the same time as providing centralized management of the entire connected environment. This is leading to some interesting new approaches to online delivery of applications, which I've discussed in my column this week on ASPnews.
Microsoft, of course, is well aware of the potential to extend desktop applications with network services, and vice-versa, as Eric Knorr discussed in an article on ZDNet last week: "Success among application servers is measured by how many enterprise applications they touch ... two-way connections to desktop data via SharePoint and web services could give [Microsoft] an edge that other app server vendors will have difficulty duplicating."
As my subject heading hints, however, there are two sides to this coin. Yes, Microsoft's desktop hegemony hands a terrific competitive advantage to its server line-up. But at the same time, the move to standardized web services architectures is diluting the effect of that advantage day by day. The Windows/Office environment used to be Microsoft's private garden. Web services mean that anyone is now free to come in and play there.
posted by Phil Wainewright 4:41 AM (GMT) | comments | link
Monday, June 03, 2002
The secret of integration is separation
"Modularization of applications, and their delivery as a service, is the biggest change that web services will create in terms of enterprise applications architectures," according to Neil Ward-Dutton, of market research company Ovum. The quote comes from a detailed review of the challenge posed to EAI vendors by web services in Infoconomy's Collaborate newsletter last week. It goes on to note that the majority of EAI vendors' existing web services strategies do not support this view of web services, putting them at a disadvantage compared to competitors that operate at a business process management level.
Ward-Dutton's point is that, before anyone can achieve effective business process integration using web services, the monolithic architectures of traditional enterprise applications have to be broken up into a modular, fully componentized form. Unfortunately for the EAI vendors, this leaves them well and truly snookered, because they have no choice but to serve the installed base of monolithic applications, where their solutions are most needed even though that market will dwindle to nothing as enterprises gradually upgrade to more modular alternatives.
JD Edwards unveiled a foretaste of those next-generation architectures last week, when it began the roll-out of the new version of its enterprise applications suite. "JD Edwards 5 is a service-oriented architecture that separates the application services from data and presentation layers," the vendor's COO Hank Bonde explained in this Infoworld article.
"The simplest example of integration is the fact that any application can now access the address book," he noted. The example is apt, because it deftly shows how integration has been achieved by completely decoupling the address book from each of the individual applications that make use of it. Only then can it become readily available to all of them.
posted by Phil Wainewright 7:02 AM (GMT) | comments | link
Assembling on-demand services to automate business, commerce, and the sharing of knowledge