Level 1: Developing the NeuroGrid infrastructure
Members of the NeuroGrid consortium have experience with globus and Grid services using OGSI compliant globus. We are aware of the migration from OGSI to the web-service resource framework (WSRF)1. The NeuroGrid framework will be: service-oriented (WSRF web service-compliant), secure (complies with relevant guidelines and laws and that it offers appropriate fine-grained access control mechanisms), flexible (the requirements for suitable applications joining the NeuroGrid framework will be minimal), scalable (NeuroGrid will be designed to work in more than the initial project six sites, and to expand to higher data throughputs), and generic (NeuroGrid will not be restricted to the initial project exemplars). NeuroGrid faces several generic e-Health challenges, and will benefit greatly from achievements within the e-DiaMoND project. To date, e-DiaMoND has established a flexible mapping between DICOM and relational databases, federation of databases, Grid-based data loading, Grid-based querying, pseudo-Grid-based data retrieval, an extensive security analysis. e-DiaMoND uses OGSA-DAI to access databases, taking advantage of the configurability of OGSA-DAI and also exploits the extensibility of OGSA-DAI’s activity system, which allows the data access component to perform additional transformations on a query result and to send that result to a separate destination. e-DiaMoND uses this extensibility to write application-specific activities, i.e., requests that perform particular operations on the data stores. This improves both the efficiency and the maintainability of the application.
The e-DiaMoND team will provide substantial expertise to NeuroGrid in a number of areas of shared interest, including Grid technology, image databases, the DICOM encoding of non-image data, and authentication and authorisation mechanisms. e-DiaMoND will also facilitate developments in several key areas:
- Intelligent replication of data.
- Dynamic access control mechanisms.
- The use of a lightweight content management system (e-DiaMoND currently utilises IBM's Content Manager).
- A workload and performance analysis to determine the most appropriate architecture.
- Application level (or semantic) query optimisation.
The main tasks will be:
- To determine the behaviour and future needs of the exemplars;
- To derive appropriate ontologies, interfaces, protocols, and standards to define the notion of NeuroGrid compliance to a federation of databases for the exemplars;
- To develop the secure NeuroGrid framework in accordance with the above ontologies, interface protocols, and standards;
- To validate the framework with respect to the exemplars.
The e-Science effort is concerned with moving from the application-specific to the generic and offering the opportunity to leverage the wealth of data captured for existing and future projects. The development of the generic framework will be in accordance with the results of the recently announced MRC / BBSRC / Wellcome Trust / NERC / DTI / JISC Joint Data Standards Study and will need to have the following functionality:
- The framework should offer federation over existing and future data sources that are NeuroGrid-compliant;
- The federation should offer the opportunity for fine-grained access control in accordance with the policies of individual applications;
- Together, this secure, federated framework should offer the opportunity for secure distributed query processing and data mining;
- The federation should ensure easy incorporation of future NeuroGrid-compliant applications.
