DODS Data Locator Service: Integrating Directory, Inventory and Data Granule Location and Access Functions

George Milkowski
Graduate School of Oceanography
University of Rhode Island
Narragansett, RI 02882-1197
george@zeno.gso.uri.edu

The Distributed Oceanographic Data System, DODS, is developing a Data Locator
service which will enable a user of DODS to locate and access scientific
research data over the World Wide Web.  The DODS Data Locator Service will
integrate location and access functions of directory, inventory and data
files within a single service through a uniform network interface.
Conceptually DODS provides an infrastructure and protocol for directly
accessing distributed data resources from a researcher's data analysis
applications. In order to navigate this infrastructure a Data Locator Service
is required which can support both data location and access.  The DODS Data
Locator Service employs an iterative locate and access procedure that allow a
user to move from the general to the detailed: (Satellite Dataset(Satellite
Pass(Region Within a Satellite Pass))) that is from directory to inventory to
data granule.  Three related data domains that are distinguished by
differences in their level of detail and semantic content are defined within
the iterative process.  These are: Datasets domain, Files domain, Records
domain, which are analogous in terms of their information content to
directory, inventory and data granule.  The Datasets and Files domains
include information that can be used to resolve a user query in addition to
pointers to objects in the domain with the next higher level of detail.  The
Records domain contains information on the structure and attributes of
individual data files that is used by DODS clients to access and interpret
the data for use by their analysis applications.  Files and Records Domain
information is maintained by DODS servers installed on individual data
archival systems.  Datasets domain information is maintained by a DODS server
installed at the the NASA Global Change Master Directory.

Beyond making it possible to locate and access data resources the network
interface of the DODS Data Locator will allow information to be exchange
between DODS servers.  This is particularly relevant to updating and
submitting new dataset references to the GCMD.  Systems that install new
DODS servers can register their dataset with the GCMD through their DODS
server and the GCMD can also access DODS served datasets and update its
references to them.