A study of business models for the federation of services supporting e-Science

Abstract

Thanks to advances in Information and Communication Technologies, scientific research has developed in an increasingly collaborative manner. At the same time, in several knowledge areas, scientific development is based on methods that require access to computer infrastructures that can perform large-scale processing as well as store and process large amounts of data. At the confluence of these two trends there is a need to integrate the computing infrastructure of research groups spread geographically to facilitate collaboration and/or increase the computing power available locally for each research group. This integration occurs through the use of service federation mechanisms that aim on the one hand enable heterogeneous computer systems to properly interoperate, and on the other hand implement business models that make the operation of the federation fair, efficient and sustainable. Much progress has been made in order to produce standards that facilitate interoperability of computer services, but advances in the business models area are still modest, with many gaps yet to be overcome. In this research project a case study driven methodological approach is followed, to propose improvements in the state of the art with regard to the definition of new business models to federate services and ensure the sustainable operation of the resulting federation. The main scientific challenge is to define business models that can guarantee fair and efficient resource allocation in federated services operated by agents who do not necessarily trust each other. Furthermore, the business model needs to be sustainable. In particular, it should avoid to the maximum the dependence of centralized services offered by third parties, or even by a specific agent among those that together operate the federation. Three federated services will be studied, each with its particularity. The first case study deals with the federation of agents that together provide a service for the processing of remote sensing images to produce ecological indicators, and is characterized as intensive in computing. The second deals with the federation of agents to provide a shared large-scale astronomical data storage service, and is characterized as intensive in data. The third case study deals with the federation of private cloud service providers, and involves different dimensions both regarding the type of computational resource involved, as well as the quality of service offered by each agent.The research methodology driven by case studies requires close interaction with operators and users of the services. This will be achieved through a series of workshops involving the team executing the project execution and these stakeholders. Twelve workshops will be held over the five years of the project's execution. Three workshops will be held to discuss the requirements of each case study. Following, eight workshops will serve to present the partial results to operators and users of each service, allowing possible realignments in the implementation of the case studies. Initially it is envisoned that these workshops will be focused on a single service, but if considered appropriate during the course of the project's execution, it is possible that some of these workshops could be merged to bring together stakeholders from various services, in an attempt to produce "cross-fertilization" in the discussions. A final joint workshop with all operators and service users of the case studies will be held to evaluate the results of the project and plan future actions. The main results expected at the end of project execution are: new business models; 3 services operating under these new models; scientific articles describing these results; training of human resources. (AU)