|
|
|
We have developed an OPeNDAP compliant, Java Servlet based RESTful web service to serve time series data. In addition to handling OPeNDAP style requests and returning standard responses, alternate output formats can be easily supported. It is also simple to reuse or customize modules to directly read various native data sources and even to perform some processing on the server.
The server is built around the Unidata Common Data Model (CDM) which merges the NetCDF, HDF, and OPeNDAP data models. The modular architecture supports pluggable Readers, Writers, and Filters via the common interface to the data, enabling a workflow that reads data from their native form, performs some processing on the server, and presents the results to the client in its preferred form. The service is currently being used operationally to serve time series data for the LASP Interactive Solar Irradiance Data Center (LISIRD) and as part of the Time Series Data Server (TSDS).
We will discuss hurdles that we have encountered with the current implementation of the CDM and present LASP's vision to broaden the applicability of a common data model. We are developing the next generation of the server, named LaTiS, that will include a fresh implementation of the data model that will continue to focus on data that are a function of time. I will also share thoughts for moving beyond the time series abstraction.
References: