HAPI: a simple, standardized interface for serving Heliophysics time-series datasets

We present a newly developed data serving mechanism called Heliophysics Application Programmer's Interface (HAPI) that can provide a single interface for requesting the digital content of time-series Heliophysics data. This mechanism has been designed collaboratively by multiple Heliophysics data providers in an effort to define a simple, least-common-denominator API that could be implemented with a minimum of effort on top of existing, diverse data services. Our initial API is focused on time-series data products, which make up a large fraction of Heliophysics data holdings. While the SPASE metadata model provides a way to describe the scientific contents of data resources, this newly developed mechanism serves as a definitive API that can be used to access the content of a data resource. There are two parts to our mechanism: a REST-like request interface (a computer-centric API defining how programs make a data request) and a data streaming format (what is the arrangement of the digital content that is the result of a request for data). Servers implementing the request interface must provide three specific capabilities: 1. list the datasets available, 2. provide metadata (mostly structural) for one dataset, 3. stream data for a given dataset for a requested time range. In terms of the data returned, two formats can be requested: a simple CSV format and a more compact binary streaming format. The details of the request interface and the streaming format will be presented. Also, several reference implementations are being developed [http://datashop.elasticbeanstalk.com; http://tsds.org], and Autoplot [http://autoplot.org] is being augmented to serve as a display client for data available through this mechanism.