Ingrid and OPeNDAP

Ingrid is used to implement the IRI Climate Data Library. There are three major goals: helping users find suitable datasets, providing a useful summary of the dataset once found, and helping users make use of the data in that dataset. INGRID uses an object-oriented approach to create a WWW-accessible environment where datasets can be found, accessed, viewed, and manipulated.

Ingrid uses OPeNDAP to exchange data with itsself, get data from other OPeNDAP servers, and send data to OPeNDAP clients. While the implementation does not depend on the OPeNDAP core, there have been no intended extensions to the OPeNDAP protocol: all the uniquely Ingrid information is encoded in the filename part of the URL, which by definition is site dependent.

The server is designed to be a WWW resource: users can write their own hypertext and simply link directly to data-based figures, calculations, even data-manipulating tools. Alternatively, datasets and arbitrary subsets of datasets can be specified as urls for web-aware tools, or can be downloaded in a variety of formats for use in older tools. While the dataset environment appears to be uniform, the datasets are spread over a network of data servers and local copies are updated automatically. The Data Library is thus both a member of and a prototype for the next generation of tools that manipulate data and metadata directly from networks of dataset servers.

There are a number of current issues that OPeNDAP community can help address:

• better data caching and transmission of update information needed for better data caching
• resolution of the conflict between clients that use name-mapping vs clients that use GRID structures: what is a server to do? Can we really set up structures to describe what is necessary to say about our data?
• standard description of data collections which are currently too large for OPeNDAP but still considered by most catalogs as a single entity (a half-century 24 member ensemble of GCM runs, six-hourly, daily, monthly, for example).
• standard description of available data manipulations