to LooselyCoupled.com homepage
 
 Weekly emails: how to advanced search
 Glossary lookup:

 

Loosely Coupled weblog


Wednesday, June 30, 2004

All the world's an SOA

With Sun and Oracle falling into line, every major systems vendor now has an SOA strategy, although most of them are more statements of intent than shipping product offerings:

  • Oracle saved its launch until last, combining it with yesterday's announcement of its acquisition of BPEL specialist Collaxa. The addition of business process capabilities allows Oracle to boast of offering "the most complete integration solution," which of course is the sort of thing the company likes to be able to do — it's never comfortable with the idea that customers might want to use its technology alongside products bought from rival vendors.
  • Sun came through a day earlier than Oracle (the venue for both vendors' announcements was Sun's own JavaOne conference, after all), but without any immediate product. Instead, it announced Project Kitty Hawk, a "comprehensive initiative" to deliver products over a two-year period commencing in the first half of 2005. The only thing customers can actually buy today is a consulting service that assesses their readiness for SOA, which is a bit rich considering that if they do happen to be ready right now, they can't buy it from Sun.
  • BEA last month revealed its Liquid Computing vision, comprising a mixture of shipping products and promised technologies such as the QuickSilver service bus and Alchemy rich client projects.
  • IBM has been talking about SOA for some time, and in April formally launched its SOA strategy with a mix of software and professional services. Like Sun, the emphasis is on helping customers start the planning process rather than moving ahead on implementations, but at least IBM does have a portfolio of products and partner offerings to draw on if a customer is eager to get going.
  • Microsoft has stopped short of formally announcing an SOA strategy but has nevertheless made it pretty clear that it is designing the next version of Windows to operate in an SOA environment. An entire section of the company's developer website is devoted to explaining SOA concepts.

This is all well and good, except that it's more a case of vendors making sure that they're "buzzword compliant" rather than rushing to embrace SOA. The main point of implementing an SOA is the ability to link to resources irrespective of whose platform they're running on, so when vendors endorse SOA, the irony is that they're implicitly endorsing the notion of platform independence. Nobody should imagine the irony is lost on them. They're well aware what the stakes are, and thus when they talk about having an SOA strategy what they're really advancing is a plan that allows them to offer SOA on their own terms. That will rarely coincide with what's best for an individual customer.

So-called 'readiness assessments' in particular are more concerned with keeping customers aligned with the vendor's own product shipment program than with accelerating generic SOA adoption. Although it's important and welcome that vendors should move their own platforms to SOA, this is just one part of the picture from a customer perspective. Unless an enterprise really does want to standardize on a single vendor's infrastructure throughout, the move to SOA has to be pursued in three separate dimensions:

  • Platform-centric: Yes, vendors do need to be pressurized to implement SOA within their product lines. But you don't have to wait for them if they're dragging their heels. Recalcitrant platforms can be wrappered or extended in order to participate in a wider SOA environment.
  • Enterprise-wide: The most important SOA within an enterprise is the one that links all its systems. If you don't have a strategy for implementing SOA across the enterprise, then any other SOA initiatives are just tinkering at the edges. Lesser tinkering can play a vital role in the wider scheme of things, but no one should be under any illusions that it's the main game. And since the vast majority of enterprises operate multiple systems (at the very least, Windows and Linux) no platform-centric SOA is likely to be big enough on its own.
  • Global: No enterprise is an island and the wider picture of SOA is that it enables businesses to interact more easily with each other. So here's another reason why platform-centric SOAs sit so far down the food chain; you can't force your customers, suppliers and partners to adopt the exact same platforms you run in-house. You have to build an SOA that will interoperate with all the other SOAs out there. Because the end game is that all the world becomes one big, multi-layered, global SOA.

PS: Apologies for late posting of this item and the one following, due to move-related systems glitches, which delayed their appearance on line for a day or two.

posted by Phil Wainewright 10:21 PM (GMT) | comments | link

Assembling on-demand services to automate business, commerce, and the sharing of knowledge

read an RSS feed from this weblog

current


archives

latest stories RSS source


Headline news RSS source


 
 


Copyright © 2002-2005, Procullux Media Ltd. All Rights Reserved.