I recently wrote a rambling, everything but the kitchen sink screed on service registries. My argument was that it doesn’t make sense to say UDDI is the only possible registry mechanism, when there are so many different service models coalescing, converging and competing at the moment. Some are legacy, some are future. ebXML, for example, has a role. Just ask Korea .
Radovan Janacek, Systinet’s VP of Engineering, came back with a response. He says he didn’t take the time to read my argument, which is fair enough. But he responds anyway, which is fair enough. I am glad he didn’t get my post though, because i don’t fully understand his response.
But i will say putting interoperability in bold every time doesn’t make an argument any more compelling.
If the answer of WS-I exponents as to “why UDDI?” is “because its good for handling WSDL” then that is fine, but a circular and somewhat tautologous argument. Isn’t that what UDDI was designed for in the first?
Luc Clement, who was on the OASIS UDDI team, but now works for Systinet, believes no vendors will offer UDDI ebXML regrep bridges (except Sun, Infravio and webMethods, who already do). As far as i can such a bridge would be an approach to interoperability, but i guess not.
What happens if RSS starts to eat some of SOAP’s lunch?
As I have said a couple of times on this blog now it would be good to talk to Systinet because it seems I don’t get it..
Repeatedly bolding a word doesn’t clear it up for me. I would be more than happy to have a corporate AR briefing on these issues. Dare i say it Radovan- I think your Corba roots are showing.
That may not be a bad thing, Corba taught the industry a lot about service integration. But every time we have tried to establish a once and future model for everything it has failed. Maybe things will be different this time.
Tim Bertrand says:
August 26, 2005 at 3:55 pm
James,
Sorry for being late to the party here, but I found a compelling need to respond. Speaking from 5 years of market expertise in selling SOA platforms/tools/standards…fill in the blank (2 years at Cape Clear and 3 years at Systinet) – I would agree that UDDI is not the only possible registry mechanism. That said, it is and has become the only broadly adopted registry mechanism/standard in customers who are seriously building standards-based SO Architectures. We constantly have customers and prospects asking about ebXML, but have yet to find more – than I can count on one hand – who have done anything with ebXML. Even in our largest customer – the United States DOD Defense Information Systems Agency (who are building common services for the entire DOD in a program called NCES) – they have found usage for UDDI *and* ebXML. They’re using UDDI (Systinet’s implementation) for anything Service related in their Service Discovery implementation. They’re using ebXML for anything related to metadata/XML/XML schema (registration). They’re going to be using Systinet Blizzard as a mechanism to bridge or integrate the two worlds.
If anyone says that UDDI is “not being implemented on an enterprise scale” or “being implemented on a departmental level” – I’d be happy to sit down and have a coffee with them and run them through 75+ scenarios in very large organizations that prove this assumption wrong.
Best regards,
Tim Bertrand
Systinet Corp.
techrainmaker.blogspot.com