Uniform Resource Names: Deploying A New Namespace Michael Mealling 19 August 1999
Michael Mealling 19 August 1999 TWIST Meeting University of California, Irvine Overview of URNs l Meant to be persistent identifiers. Binding to resource is permanent. l Simple identification is useful (XML namespace identification) l A resolution system is available if needed. l Allows for variety in the syntax of the namespace. l Examples: – URN:ISBN: X – URN:PDI://oma.eop.gov.us/1999/8/5/14.text.2
Michael Mealling 19 August 1999 TWIST Meeting University of California, Irvine Deployment Issues l Problem identification l Experimentation l Development l Adoption l Scaling l Management
Michael Mealling 19 August 1999 TWIST Meeting University of California, Irvine Problem Identification l We know the general class of problems. Now we need to look at specific ones and apply the proposed solution. – Bibliographic identifies (ISBN, ISSN, DOI) – Schema identification (XML Namespaces) – Document Management (PDI, IETF) – Distributed object systems (Globe)
Michael Mealling 19 August 1999 TWIST Meeting University of California, Irvine Experimentation l Once the problems are identified internal use and experimentation begins. – Hidden deployment issues appear. (I.e., the ‘u’ flag issue) – Reusable code begins to appear. – Corporate resource begin to be committed. – We are nearing the end of this phase. Several experiments are ready for “prime time”
Michael Mealling 19 August 1999 TWIST Meeting University of California, Irvine Development l Developers begin work on solutions to the problems. Issues to be aware of: – Interoperability – Adherence to the standard. If a large enough segment of the user base does something incorrectly, that incorrect behavior becomes a de facto standard. – Open source software provides a common reference implementation.
Michael Mealling 19 August 1999 TWIST Meeting University of California, Irvine Adoption l Moving URNs into the main stream – Must be a compelling reason for either the user or the service provider? – For the user it must be either ‘sexy’ or extremely useful (or at least a functional requirement of something that is). – For the service provider it must be essential for some product that is sellable. – Chicken and egg problem (No one uses it because the software doesn’t support it. The software doesn’t support it because no one uses it.) – Killer app?
Michael Mealling 19 August 1999 TWIST Meeting University of California, Irvine Scaling l Scaling of assignment – The IANA is currently the NID authority. Will this scale? l Scaling of resolution – Is a DNS based RDS capable of scaling well enough? l Scaling of actual resolution servers – Is THTTP to simple? – Are other protocols or protocol profiles needed?
Michael Mealling 19 August 1999 TWIST Meeting University of California, Irvine Management l To early to tell. If we knew what they would be we could fix them now. l Some possibilities: – Namespace specific knowledge allows for special casing certain URNs which affects interoperability – What happens if we a namespace becomes so popular that it “takes over” the URN space? (I.e. the.com phenomenon happens to URNs)
Michael Mealling 19 August 1999 TWIST Meeting University of California, Irvine Current Status l URN Working Group waiting on final documents which are in Last Call. l Namespace registration procedures are being setup now. At least 4 namespaces are in the process of registering. l Software is being released for browsers Real Soon Now ™.