Over 100 delegates gathered in Dublin, Ireland, on 22 and 23 May for the e-Infrastructure Reflection Group (e-IRG) bi-annual workshop. Representatives from 23 countries met to deliberate on the best policies for coordination of a European-wide open and innovative e-infrastructure.
During the welcome address, Sverker Holmgren, current chair of the e-IRG, introduced some of the highlights from their latest strategy white paper. While the internet provides a common user interface and access to common services and networks, this is not the case with e-infrastructures: integration is needed. Holmgren emphasises that users will need to become more directly involved, and be both prepared and empowered to pay for the services.
The two-day e-IRG workshop also provided an ideal opportunity to exchange best practices. Communities considered their shared aims, including developed services (e.g., storing/archiving, computational services, web services, etc.) that could potentially be adopted by other disciplines. Nuria Bel, a professor in natural language processing, described how her institution (Institut Universitari de Lingüística Aplicada, Spain) has developed over 40 services, including text-cleaning and annotation services (see video). Other services outlined in the meeting can also be found on the e-IRG website.
While it is vital at workshops such as this to find those subjects that are common to the different actors involved, areas of major divergence do also exist. This was especially evident in the Data Interoperability panel session. “Communities of data providers exist in different states of preparation, from operational distributed interoperable systems to communities not yet interested in sharing data,” explains Francoise Genova, director at the Centre de Donnees astronomiques de Strasbourg (CDS), France, which develops services for the international astronomy community. “Therefore, building an e-infrastructure will have to be a global endeavour accommodating diversity and data providers from many different contexts.”
Alberto Michelini explains that there may not yet be enough consciousness of the scientific problems that can actually be addressed by data sharing. Michelini is a seismologist and representative from the European Plate Observatory (EPOS), which has a long-term infrastructure plan to integrate data from 134 institutes, so as to better understand earthquakes and other Earth dynamic systems. Furthermore, he argues that a whole new type of scientist, targeting multi-disciplinary problems, is needed.
Bob Jones, head of EC-funded CERN IT projects, also presented the latest results from the Helix Nebula project, which is currently being piloted at CERN, the European Molecular Biology Laboratory (EMBL), and the European Space Agency (ESA). This hybrid cloud computing model aims to integrate the resources of commercial cloud service suppliers with public services, such as the pan-European data network, GEANT, and European Grid Infrastructure (EGI). “The market of end-users must be well understood by funding agencies and e-infrastructure service providers, in order to gauge impact,” explains Jones. EIROforum comprises a partnership of eight international research organisations and representatives from each partner will work together to discover where there are common needs and divergence. “While e-IRG, EGI, GEANT, etc. are looking at a slice of the picture, complete service integration is difficult to achieve from these perspectives alone. We need to know how do we use the grids, networks and the data services together, and this type of forum can give an opinion across disciplines for providers and funding agencies,” he adds. Large research cluster projects, as well as new flagship projects (e.g., the Human Brain Project, Graphene), and national research bodies, will also be represented.
During his keynote speech, Richard Kenway, former chair of the Scientific Steering Committee of the Partnership for Advanced Computing in Europe (PRACE), again emphasised the users’ responsibility. Technology is rapidly advancing and it is therefore largely unpredictable what new capabilities to anticipate with the coming data deluge, exascale computing, ‘internet of things’, etc. The science is also evolving rapidly, as are the problems that we are attempting to tackle. Computer models will develop, but these are not necessarily commensurate with the potential importance of their result (i.e., socio-economic grand challenges). Kenway also quoted Robert Madelin, the European Commission’s director general for communications networks, content and technology: "If we had tomorrow's e-infrastructure capability, what bad things could be avoided". In addition, Kenway calls for there to be less of a 'technical push' and more of a 'user pull' with communities self-organising to develop their value proposition − a co-design approach. Horizon 2020 is an invitation for the user community to design centres of excellence that suit them, says Kenway. He also predicts that a truly competitive e-infrastructure will need an accepted value proposition and a skilled workforce to maximize impact. He argues that lessons can be learnt from CERN in building real partnerships between the users and the technical providers.