Presentation on theme: "Introduction to FAIR-TRADE Todd King 1, Raymond Walker 1,2, Lee Bargatze 1, Deborah McGuinness 3, Everett Toews 4, John Shillington 4, Robert Bentley 5,"— Presentation transcript:
Introduction to FAIR-TRADE Todd King 1, Raymond Walker 1,2, Lee Bargatze 1, Deborah McGuinness 3, Everett Toews 4, John Shillington 4, Robert Bentley 5, Tomo Hori 6, Robert Rankin 4,7 1 Institute of Geophysics and Planetary Physics, UCLA 2 Earth and Space Sciences Department, UCLA 3 Dept. of Computer Science and Cognitive Science Department, Rensselaer Polytechnic Institute 4 Cybera, University of Alberta, Edmonton 5 Dept. of Space and Climate Physics, University College London 6 STE laboratory, Nagoya University, Japan 7 Dept. of Physics, University of Alberta Edmonton
What is FAIR-TRADE A proposed project to NSF ◦ NSF : Software Infrastructure for Sustained Innovation (SI2) A framework to integrate all resources in a (distributed) research environment. ◦ FAIR: Flexible Application of Informatics in Research An implementation of computational resource sharing ◦ TRADE: Transparent Resource Allocation and Delivery Environment
Why FAIR-TRADE? Enable a more resilient, open exchange of data and computational resources. Mitigate the logistically difficulty with federated data storage. Leverage existing computational resources. Abstraction of the entire research enterprise. - Foremost - Transition to an objective oriented model for research.
FAIR-TRADE Concepts All resources (data and computation) have unique identifiers Each resource has an appropriate set of attributes. Reasoning is performed to match resources as needed. Example: Apply an application to data. In a FAIR-TRADE system the platform to run the application will be selected based on available resources, volume of data, proximity to user and data, bandwidth, cost and other factors.
Data Computation Data Model 1 Data Model 2 Environment 1 Environment 2 FAIR-TRADE Task Conceptual model of the FAIR-TRADE framework showing data from two sources each with a different data model and two computational environments. Tasks are submitted to the FAIR-TRADE framework by users. The framework will locate, retrieve, and join the resources to achieve the prescribed task.
Functional View – For Digital Resources Locator: A search engine that can locate resources based on a set of constraints. Returned results always include the unique resource identifier, plus any defined attributes. Resolver: Given a unique resource identifier, locate and retrieve the full complement of metadata associated with the resource. Retriever: Given a unique resource identifier and optionally the desired time span and format, retrieve the digital resource. Registrant: Stores an instance of a resource, which consists of structured metadata that may point to a physical storage location. Returns the unique identifier associated with the resource. Repository: A location for the physical storage of digital resources. Stored resources are accessible by standard protocols such as HTTP, FTP, and secure copy (scp). Analytics: A collection point of metrics from each service. Services will report activities to the service.
Functional View – Computational Resources Controller: Initiates tasks by locating the appropriate computational environment, spawns a monitor for the task set, and invokes a launcher for each task. Launcher: Prepare the environment for the initiation of an application, which may include the retrieval and localization of digital resources. Monitor: Provide task supervision and user notification.
Function View – Big Picture UserControllerRetrieverMonitorApplication Task Set Initiate Launcher Resource ID Resolver Metadata Registry Register Monitor Resource ID Metadata Publish Run Status Results Post Report
Next Steps SPASE is a good starting point for defining models for computational resources. Current Resource Types (Classes) AnnotationNumerical Data CatalogObservatory Display DataPerson DocumentRegistry GranuleRepository InstrumentService Suggestion is to add: ◦ Application ◦ Platform