There currently exists an automated procedure for registering URNs for online dissertations. The aim is to make it possible to register URNs at the German National Library as far as possible with no additional work. For this reason, the registering process of URNs is being linked to the procedure for registering online dissertations. This has been implemented by extracting and then registering the URN and the associated URL from the "MetaDiss" metadata set. The URN/URL pairings are registered however generated after the online dissertation is bibliographically indexed and archived by the German National Library. The following properties characterise this procedure:
A variety of factors influences the technical stability of the service, such as the hardware platform used,
hardware data-redundancy and recovery systems in case of system failure. An aspect that calls for particularly
careful consideration is the database management system being used. The prototype is based on MySQL. As regards
data consistency, quick recovery systems in case of a database failure and synchronising data with external
databases however, this is not adequate to provide a productive service. There is a need to migrate the service
to a database management systems that will meet these requirements.
A productive system implies the need to take scalability of the service into consideration. From this stems the need to construct a mirror.
In order on the one hand to be able to reduce the time lag between generating and registering a URN and on the other hand to be able to offer a universal interface for managing URN/URLs associated data, the link between this interface and existing business procedures has been cut.
The functions described will be implemented using an XML based data transfer format. All URN/URL registrations received at the German National Library, for example by e-mail or web form, will be converted into this format. Subsequent, internal, automated processing of registrations can thus be carried out. At the same time, new delivery methods can be incorporated without difficulty.
The following delivery methods are currently envisaged:
URN/URL data supplied via these various delivery methods, such as email or OAI, will be converted into the XML based data
transfer format. The XML based data structure will be internally transferred to an "Ingest" module with modular
programming. There the consistency and completeness of the registration will be checked as well as checking for duplicate
URNs/URLs. To ensure consistent data, on the enhanced data model duplicates will also be blocked at the Sybase database
level. Once the registration has been successfully processed, and should erroneous data be found, then the institution
concerned will be notified.
To enhance service availability, work will start in early 2004 on setting up an external mirror at the Bibliothekszentrum Baden-Württemberg (BSZ). Data exchange will occur at protocol level, not at database level. To achieve this, an XML based interface will be implemented at the German National Library and the Bibliothekszentrum Baden-Württemberg. Data comparisons at broadly spaced intervals are planned for the opening phase.