Keep in mind that the *primary* purpose of a registry is to enable dynamic information exchange among heterogeneous distributed infrastructure products (platforms, mediation systems, and management systems). Non-human driven information exchange requires predefined data models and semantic behaviors. I just don't see how a "writable web site" can replace a standard registry data model and protocol.
There are limitiations to the UDDI data model and it's associated protocols. And architecturally, I think a resource-oriented API might be preferable to an operation-oriented API. (Each entity in a UDDI registry has a unique identifier, so the model could be adapted to support a REST interface). But -- the registry data model and the protocol must be standardized to enable information exchange. So first someone has to define a new data model, and then they would have to define the semantics of the REST API. And then they would have to usher it through a standards body. So, I'd say that's at least 4 years before it's usable -- that's assuming that someone is initiating such a project. I've heard rumors that IBM wants to push its new WSRR data model and protocol into a standards body. We'll see. (btw -- WSRR's API is not RESTful. Systinet's repository API is RESTful, and it is extremely powerful. But that's a repository, not a registry.) Anne On 2/7/07, Sanjiva Weerawarana <[EMAIL PROTECTED]> wrote:
Hi Anne, Are you saying that UDDI will (still) become the de facto registry access protocol? I'm more orienting towards the evolution of an LDAP-like lightweight alternative (offering primarily a REST interface .. in other words, a writable Web site for the most part) to become the de facto standard. Sanjiva. On Tue, 2007-02-06 at 11:47 -0800, Anne Thomas Manes wrote: > Note that the report I wrote is not just about UDDI -- it's about > registry, repository, and governance. UDDI is just one small aspect of > the greater set of governance systems. But as this interview > indicates, I view support for the UDDI protocol as an essential > requirement (but not the only requirement) for a runtime registry. > > Anne > > On 2/6/07, Gervas Douglas <[EMAIL PROTECTED]<gervas.douglas%40gmail.com>> wrote: > The original definition of Web services included SOAP, WSDL > and UDDI, but the latter was often ignored, until now, as a > Burton Group report says UDDI v3.0 is emerging as a key > standard for SOA registry and repository technology. > > The importance of the UDDI standard in the future of SOA is > highlighted in a new Burton Group Inc. report, "Registry > Services: The Foundation for SOA Governance" by Anne Thomas > Manes, research director at the analyst firm. In this > interview, Manes explains why after being ignored for so long, > the OASIS UDDI standard now at version 3.0, is finally moving > up the adoption curve. > > > > > > > > > > > > > > > > > > > > The original definition of Web services was that it was SOAP, > WSDL and UDDI, but nobody ever seemed to include UDDI. Why has > UDDI lagged in adoption for Web services and SOA? > Anne Thomas Manes: That's because UDDI was part of the > management space. You never need management right at the > start. At the beginning you need the development tools. That's > the core. SOAP and WSDL gave you the core development tools to > go out and build Web services. You don't have to do management > until you have systems that are running in production. That's > why UDDI is slowly, but surely gaining traction. I think it's > a lot more accepted now than it was a year ago. > > Do you see drivers now that might speed the adoption of UDDI > for SOA implementations? > Manes: Staring in 2004 the innovators were adopting UDDI. In > 2005, it was the early adopters. And in 2007, we might cross > the divide and get to the early majority. > > Is that because UDDI has become necessary for SOA? > Manes: I think so. You don't need UDDI to get started with Web > services. You don't need UDDI to enable integration among > applications. But if you want to do SOA, you have to start > managing the environment and UDDI becomes the system that > enables communication among multiple environments. UDDI is the > foundation for governance. As people start deploying more and > more services and their systems get further and further out of > control, they realize that they need to do something. And they > start by bringing in a registry. > > So is the need for a registry driving UDDI adoption? > Manes: Usually, they figure out pretty soon that a registry is > not enough and then they have to bring in a repository and > start contract management and policy management, but it's > really only the innovators who have reached the true > understanding of the meaning of governance. > > Why is UDDI important to the registry in SOA implementations? > Manes: The true value of UDDI is not for discovery of > services. It's not like a developer uses UDDI to figure out > where a service is. The purpose of UDDI is for the various > components of your runtime infrastructure to be able to share > information about services and dependencies and policies that > apply to the services that are out there. So the value of UDDI > is that it's a standard protocol to talk to a registry. The > registry provides this information exchange. If I don't know > how to talk to the registry, I can't get that information. So > the protocol to talk to the registry is UDDI. It's a critical > component of the system. > > Are there any competing technologies? > Manes: Well IBM has created a whole new API. It's called IBM > WebSphere Registry and Repository and it's been shipping for > about six months. So it's possible that IBM is going to turn > around and say, "We've created a whole new format and > everybody else should adopt our approach." Is everybody in the > world is going to jump on the bandwagon and do it the way IBM > says? I don't know about that. > > > > Now, isn't there an issue as to whether to use UDDI or ebXML? > Manes: I don't think there's an issue at all. There's a spec > out there called ebXML Registry, but nobody's using it. > > What do you see as UDDI's strength as a standard, versus other > proposals such as ebXML Registry or IBM's WebSphere Registry > and Repository? > Manes: The problem is this: if I use AmberPoint for > management, and I have Sonic ESB, and I have the Reactivity > XML Gateway, and I'm still building services with WebSphere > and .NET and Ruby on Rails, how do all those systems > communicate with IBM's registry? It doesn't work. If I throw > in the Systinet registry or the Infravio registry they all > know how to talk to UDDI. They can all share the information. > > So will this be the year we see some real forward movement for > UDDI? > Manes: Certainly, I've seen steady increase in interest in > UDDI over the last two to three years. It's slowly gaining > adoption. > > You can read this interview at: > > http://searchwebservices.techtarget.com/regActivateSiteMO/1,296514,sid26,00.html?NextURL=http%3A%2F%2Fsearchwebservices%2Etechtarget%2Ecom%2ForiginalContent%2F0%2C289142%2Csid26%5Fgci1230185%2C00%2Ehtml%3FOffer%3DWSintesb1211&priTopic=299051 > > Gervas > > > > > > -- Sanjiva Weerawarana, Ph.D. Founder & Director; Lanka Software Foundation; http://www.opensource.lk/ Founder, Chairman & CEO; WSO2, Inc.; http://www.wso2.com/ Director; Open Source Initiative; http://www.opensource.org/ Member; Apache Software Foundation; http://www.apache.org/ Visiting Lecturer; University of Moratuwa; http://www.cse.mrt.ac.lk/
