[Date Prev] | [Thread Prev] | [Thread Next] | [Date Next] -- [Date Index] | [Thread Index] | [Elist Home]
Subject: FW: [regrep] RE: [uddi-spec] ebXML Registry / UDDI Convergence
Posting email which bounced from previous send. Christina Portillo Advanced Computing Technologist Data Management Services - Architecture & eBusiness The Boeing Company Phone: 206-544-7474 PO Box 3707 M/S 2R-91 Fax: 206- 544-5889 Seattle, WA 98124-2207 christina.portillo@boeing.com ----------------------- xml.web.boeing.com xml.ca.boeing.com ebusiness on the web = XML + JAVA + Web + Controlled Vocabularies -----Original Message----- From: Portillo, Christina Sent: Friday, September 27, 2002 11:57 AM To: 'uddi-spec@lists.oasis-open.org'; 'regrep@lists.oasis-open.org' Cc: Breininger, Kathryn R; 'bellwood@us.ibm.com'; 'lclement@microsoft.com'; 'mmealling@verisignlabs.com' Subject: RE: [regrep] RE: [uddi-spec] ebXML Registry / UDDI Convergence Respectfully submitted to the chairs of the UDDI and ebXML Registries: I have sent this note to both list and both chairs of the UDDI and ebXML registries in case my posting to your lists fails. Please post this to your lists for discussion if I do not have permission to post. Thank you. DISCLAIMER: The thoughts below represent the current opinions of a small team of architects focused on the larger eBusiness data sharing collaboration needs. All the work and thoughts I reference are still very much in a draft state and total consensus on these notions have not been achieved in the architecture community. Proposal I support the convergence of the ebXML and UDDI registry information model. I see the need for the convergence of the IETF ideas (Michael Melling mmealling@verisignlabs.com ) for a namespace registry information model into this mix also. <http://www.ietf.org/internet-drafts/draft-mealling-iana-xmlns-registry-04.t xt> One registry model Discussion Boeing architects are in the process of trying to find a way to make the ebXML and UDDI model work to support web services and application (conversion/Translation) processing work. I have been pulling together functionality lists of the 2 specifications trying to see where the gaps and overlaps are in regard to meeting the Data Sharing requirements of Boeing. <<ebxml-UDDI-Comparison-020911.doc>> As you can see from the comparison list attached there are gaps in both registry models and services which do not support the full range of required functionality to support our eBusiness Collaboration model. At a lower level the UDDI specification has addressed the need to register the artifacts which support the management and use of 1 or more web services within a domain or partition. At a slightly higher level the ebXML Reg/Rep specification has addressed the need to register artifacts which support transaction processing. There is a need for a higher level focus on the uses of a registry to accommodate the thinking coming out of the UNCEFACT TWG. A business, enterprise, or consortium may be owners of one or more domains. A domain may represent some area in the Business Operational View of a business or a set of collaborating businesses. The UNCEFACT TWG (old EBTWG old ebXML BP teams) see the importance of defining and registering the artifacts that support the definition and model of a business domain which may have one or more collaborations specified. All of the ebXML specifications may generate a whole host of artifacts which need to be shared in some public, private, or semi-private registry. See slides 9-13, & 16 in the presentation David Welsh made to the NW EC/EDI Roundtable in September http://www.ecnw.org/. <<Northwest-EC-Roundtable-Presentation-Sept-24.ppt>> We see the importance of registering information which would support the partitioning, management, coordination, and evolution of a domain in respect to the private and public aspects of this business process. Based on all these needs a small team or architects sees the need for a registry model which accommodates registration of: 1. artifacts that support/define the modeling, semantics, sequencing, collaboration of a business collaboration 2. artifacts that support/define the transactions and messaging 3. artifacts that define resources needed to support web services (ala WSDL) 4. artifacts that define resources needed to support application/namespace processing (ala RDDL). 5. namespace name (ala IANA or DNS) and have these registered namespace names associated to documentation or other files which support all of the above. 6. artifacts that support/define the relationships of those organizations which have a dependency on the artifacts in 1 & 2. 7. artifacts that support/define the events to which these organizations have a dependency in the business collaborations. categorization schemes (ala REA and UDEF) which can help in classification, relationship mapping, and dependency mapping of these many private, semi-private, and public artifacts to a single domain or specification. Aids to tying metrics or oversight to these. 8. metrics on performance of services, models, collaborations, .....to gauge the efficiency of the specifications in support the management and further evolution of these specifications. 9. network navigation schemes which make it easy to allow specialized views of this information which is independent of the hardcoded information model of the builders of the registry (needed to do federation) which might be satisfied with pursuit of topic map enabling in registries. We would like to see a registry model where all of these artifacts would be stored in the registry with information model strategies between these registries working as a single registry information model if not in fact becoming a single registry architecture. I would like the appropriate metadata elements of the WSDL and RDDL folded into the RIM. However, we still see the need for the continuation specification of sets of Registry Services targeted to specific audience such as: * web service interfaces between systems, * business collaboration Modeling and analysis, * business collaboration choreography & management, * Namespace registration and, * Domain (partitioning) ownership registration. <<ebXML-UDDI-Information-Models-020911.ppt>> The information model should be able to accommodate new thinking for resource discovery coming out of the Semantic Web effort. We would like the RIM and Registry services to be able to take advantage of the functionality of navigation and discovery being specified in the Topic maps and RDF specification accommodated into the registry architectures so all of these could be architected to work on the same sets of information. We would like to have a common resource discovery model that would support services, applications, architects, business owners, customers, suppliers, and partners in the need to analyze information. Namespace Discussion We suspect that we will have hundreds or thousands of namespaces specified which may support one web service or a whole community of users (we call them domains). We need a way to keep the user specific and application specific uses of the namespace as a qualifying name independent of the uses of the namespace as a categorization mechanism to establish the semantic and modeling definitions of the set of information used by a domain to support data interchange/sharing. The name of the namespace should be registered. It should not point to any file. Files or records should point to it as a categorization mechanism which may support different services in different ways. "URI is a stable and a unique identifier. Once assigned it should not be reused or altered in any way. People and applications need to find out information about the agreements and specifications which went behind the thought of establishing this namespace. People could find the necessary information about a namespace through location of information assets records and their referenced objects registered in the registry. Location of these does not require that the URI be resolved down to a URL or to a RDDL file. It requires that standardized search criteria (metadata) be established to locate the namespace via its domain association, domain type, categorization terms, or its unique name. A namespace record gets handled pretty much like registration of a domain. The actual URL this namespace resolves to if required could be handled via an indirection table (alias table) much like the identifier http://xml.ca.boeing.com is resolved down to http://bruin.ca.boeing.com/xml/ is resolved down to http://nt-img-01.ca.boeing.com/xml/ which is further resolved down to 132.224.203.174. There are tools and services in place that run in the background which do the resolution. I question whether we are setting ourselves up for problems by having the URI be used as a hardcoded URL which resolves down to a single file. A registry can provide some of the support of a DNS like registry for uniquely naming/registering a namespace. But there are other tools that need to be put in place which do the heavy lifting of the namespace resolution down to a resource file or resource." The namespace registry indirection table should point down to the resources which: * provide the models, worksheets, collaborations, patterns, glossaries providing semantic understanding * provide the resources needed by a web service (ala WSDL) * provide the resources needed by an application (ala RDDL) * provide the resources needed for navigation (topic map or RDF) * identify what belongs to that namespace * what has a dependency on that namespace * what metrics have been gathered about the functioning of that namespace. These in turn point back to their association to a namespace. Hence I think what is needed in the registries is an architecture, a concept of an indirection table, to help services or applications resolve a namespace down to specific set of resources needed to support that application or service whatever those might be. A namespace should be able to support multiple application specific or service specific RDDL or WSDL files all of which play nicely in that namespace or multiple namespaces. The ownership and the overall specification of the namespace boundaries would be under the configuration control of the owner but would not tie the hands of the users of the namespace to tightly to implement (or extend) the specified structures given their different toolsets and platform. <<Services-Reg-Rep-020829.ppt>> <<Namespaces-Reg-Rep-020813.ppt>> Christina Portillo Advanced Computing Technologist Data Management Services - Architecture & eBusiness The Boeing Company Phone: 206-544-7474 PO Box 3707 M/S 2R-91 Fax: 206- 544-5889 Seattle, WA 98124-2207 christina.portillo@boeing.com ----------------------- xml.web.boeing.com xml.ca.boeing.com ebusiness on the web = XML + JAVA + Web + Controlled Vocabularies
Attachment:
ebxml-UDDI-Comparison-020911.doc
Description: MS-Word document
Attachment:
Northwest-EC-Roundtable-Presentation-Sept-24.ppt
Description: MS-Powerpoint presentation
Attachment:
ebXML-UDDI-Information-Models-020911.ppt
Description: MS-Powerpoint presentation
Attachment:
Services-Reg-Rep-020829.ppt
Description: MS-Powerpoint presentation
Attachment:
Namespaces-Reg-Rep-020813.ppt
Description: MS-Powerpoint presentation
[Date Prev] | [Thread Prev] | [Thread Next] | [Date Next] -- [Date Index] | [Thread Index] | [Elist Home]
Powered by eList eXpress LLC