The Visualization Workflow Pipeline prototype consists of two sub projects led by the University of Washington.  A third project led by the University of Washington is Unstructured Grid Services.

Two Objectives

1) Design a new RESTful API to Trident and exercise it using COVE (COVE + Trident)
2) Demonstrate that data stored in Azure can be accessed by COVE (COVE + Azure)

 Summary of Deliverables

1) COVE + Trident: Demonstration of interoperabilty between COVE and Trident using a new web services API to access and manage the workflow system, decoupling Trident from the local desktop for easier integration within the OOI.  The WS API specification, code, and appropriate documentation will be made available to OOI engineers via the Confluence website.

2) COVE + Azure: Demonstration that any OOI application, exemplified by COVE, can access and manipulate data managed in the Microsoft Azure cloud.  Code to facilitate access, including workflows used to process data will be made available to OOI engineers via the OOI Confluence website.
 
 Technologies

1) Trident is a workflow and provenance management system, with storage layers for RDBMS, Amazon S3, and local files.  Trident is applicable to requirements for complex analytics in the OOI.  A RESful API to Trident provides interoperability with the rest of the OOI framework.

2) COVE is a visualization system providing a richer UI model and better support for oceanographic data that Google Ocean.  COVE serves as a "canonical application" for visualization and management of ocean data.  That is, demonstrations with COVE provide a proof of concept that other client-side systems can successfully interoperate with Trident and Azure.

3) Azure is a new cloud computing platform provided by Microsoft.  Azure is a platform for massively parallel asynchronous computation.  We are exploring the role that Azure plays in the OOI by "stress-testing" the APIs.

 Architecture
 
Status

Both the Windows and Mac COVE user clients are feature complete at this point.  The client provides the ability to create geo-positioned visualizations of scientific data, arbitrary bathymetry, images, and instrument layout using a Google Earth - like zooming interface and data layering approach.

The COVE visualization engine is available both on the users system through the client or as a visualization service through a RESTful web service.  This allows any visualization created with the client to be migrated to the web to be regenerated automatically as datasets change.

The COVE data store is implemented as a RESTful web that is accessible through the COVE client. The data store can be located on a user machine, the web, or in the cloud using the AZURE interface.  XML based metadata is stored with each uploaded item to provide search parameters.

The COVE client provides scientific workflow support through the Trident Workflow Web Service.  An integrated interface allows the user to access workflow output from the client.  A base activity set has been implemented and several sample activities and workflows have been created to exercise the activities.  The activities have been architected to enable data access in the cloud through URI's.

TODO

Complete COVE data server web interface

Improve UI for migrating visualizations to the web for scheduled regeneration.

Investigate ability to run COVE visualization engine on cloud services.