find your perfect postgrad program
Search our Database of 30,000 Courses
Computing, Science and IT Research: Broadening e-science’s horizons
Research is rarely about working in isolation. Far more is discovered when a group of specialists work together. Interoperable grid software makes such co-operation not only easier and faster, but also more productive and successful, as Sara Goodwins discovers
As science becomes more complex and scientists increasingly need to handle vast amounts of information, so grid computing is gaining in importance among researchers. A single computer often lacks sufficient capacity to deal with the volume and complexity of scientific work, but the combined capacity of a number of computers can handle much more.
For over a decade, users have been able to combine computers in clusters to increase their capacity. However, such clusters have usually been restricted to particular organisations that share common software. Grid computing links computers over a much wider area, often outside national and regional boundaries. All those working on a particular project, whether in London, Latvia or Los Angeles, will be part of the appropriate grid, and will share data, transfer information and utilise the far larger computing capacity.
Even though grids have hugely increased the computing capacity of e-scientists, they can be self-limiting. Each grid relies on its middleware to organise and integrate its resources, and that middleware may not be able to talk to and co-operate with the middleware of other grids. Researchers are therefore unable to work easily with colleagues using different grid platforms. As a result, new ways to connect heterogeneous grids were needed to remove the isolation.
Improving communication
A solution to the problem of communication between grids was considered as early as January 2004, when the University of Southampton founded the Open Middleware Infrastructure Institute (OMII). Two years later, OMII-UK was formed, when the Universities of Southampton, Manchester and Edinburgh began their collaboration to provide free, open-source software to help e-researchers. In May 2006, OMII-Europe was developed out of OMII-UK.
Promoted as one of the European Union’s (EU) four flagship projects, OMII-Europe was established to provide key software components for building e-infrastructures within the European Research Area (ERA). The project ran for two years from May 2006 to April 2008 and received a grant of 5 million from the EU. Steve Brewer, Deputy Project Manager at the University of Southampton, says: ‘The main goal of the EU was to develop accessible infrastructure across the whole of Europe, while at the same time disseminating European knowledge as widely as possible.’
The project was proposed and co-ordinated by the School of Electronics and Computer Science (ECS) at Southampton University and involved 16 partner organisations worldwide. Eight were from Europe, and included the Universities of Southampton and Edinburgh as well as research establishments in Germany, Sweden, Italy and Poland. In addition, four of the partners came from the US, including the Universities of Chicago and Illinois, and four were from China, including Beihang University and the China Institute of Computing Technology. Steve Brewer comments: ‘The project was multidisciplinary and the partners were chosen for their breadth of experience. We also wanted to ensure that the team was as objective as possible and not allied to any particular middleware.’
|
Middleware – what is it? Grid computing refers to the presentation of dispersed and diverse computing resources as a unified service. Grids are a major contributing factor to the work of e-scientists, but would not be possible without specialist software called middleware. |
Such was the nature of the international research team that many nationalities were involved, not only those from the countries whose organisations were directly involved. Although pan-European in scope, the project was limited to two years’ duration, which made it an ideal length for postgraduate students and researchers from outside Europe. They could get involved in a project with a specified end date that matched the length of their postgraduate programme and didn’t exceed their funding.
Steve again: ‘The meetings were particularly stimulating, driven as they were by the different expertise and perspectives which the various researchers brought to the project.’
International scope
The aim of the project was to remove some of the barriers to e-science and so increase the possibilities of international co-operation. The OMII-Europe team wanted to enable researchers to cross national, European and global boundaries, and use systems to which they would not otherwise have had access. The emphasis was on re-engineering the software components rather than on developing new technology, and thus on extending rather than replacing the interface.
Two alternative solutions were possible for enabling different grids to work with each other, ie to be interoperable:
- use adaptors to translate specific design aspects from one domain to another, or
- use the native ability of grid middleware to interact directly via well-defined interfaces and common open standards.
The creation and implementation of common standards was considered the better alternative, particularly when combined with the development of standards on all middleware platforms. A single standard means interoperability and that benefits:
- grid developers, as a standard set of services makes applications portable across different grid middleware systems
- e-scientists, as they potentially have access to a significantly larger set of resources and common ways of accessing them
- e-resource owners, as using only one middleware system reduces management overheads.
OMII-Europe therefore focused on developing a common set of application-level services and working with a number of grid middleware platforms to encourage their adoption. Much work on common standards for grid computing had already been done by the Open Grid Forum (OGF). An international community of users, developers and vendors, the OGF is developing standards and specifications for use on distributed systems. OMII-Europe used OGF standards for their software components, and several of the OMII-Europe project developers are members of OGF working groups. As Steve Brewer says: ‘The knowledge and enthusiasm of the team are re-emerging in future projects.’
One of those projects is the EU-IndiaGrid, the first collaborative e-science grid project between Europe and India. Seamless interoperability is crucial to successful collaboration between the European grid infrastructures EGEE (Enabling Grids for E-sciencE) and DEISA (Distributed European Infrastructure for Supercomputing Applications), and India’s National Grid Computing Initiative, GARUDA. EU-IndiaGrid e-scientists wanted to collaborate on work on quantum atomistic simulations, which are very demanding in terms of processing and memory requirements. OMII-Europe components were used to ensure interoperability, e-scientists were able to use all three infrastructures, and the ease and speed of work was increased.
Achievements
Open standards make software compatible, open sources make it available. OMII-Europe has re-engineered a number of components that allow identically specified jobs to be run, managed and migrated to different existing middleware platforms. Creating the standards was only the first step. OMII-Europe needed to encourage their adoption and use by ensuring that e-scientists knew how to use them.
Training was a significant part of the OMII-Europe project. The training team had bases in Europe and China, and not only conducted training on the OMII-Europe components, but also various grid middleware platforms, so that users could fully understand the components deployed on them. OMII-Europe aimed to establish itself as the provider of impartial advice about e-infrastructures, and has also made the project’s components and supporting information freely available in its online repository.
Common interoperable services are database access, virtual organisation membership, accounting, job submission and job monitoring. Many of the major grid software providers worked with OMII-Europe to achieve compatibility with their own e-infrastructure, and forthcoming versions of some of Europe’s main grid platforms will incorporate software developed by OMII-Europe. Researchers will therefore be able to access grid computing resources with the same ease that they can access information from the Internet.
Who uses it?
The project’s components, downloaded from OMII-Europe’s online repository, are being adopted by European research initiatives like WISDOM (Worldwide In Silico Docking On Malaria), which is focused on developing drugs for malaria and other neglected and emerging diseases, including avian flu. Malaria affects 300 million people in the world. One million people die of it every year and one child dies every 30 seconds. Through grid computing, the WISDOM project shares r 
esources across continents, collects and disseminates information, routinely mobilises resources, and performs CPU-intensive tasks. The project, through its federation of databases, collects biomedical data and improves disease monitoring. However, WISDOM was hampered by the inability of different grids to talk to each other. The willingness to share information existed, but the e-infrastructure did not.
Installing OMII-Europe’s interoperable software created the vehicle for cross-organisational collaboration. Dr Alistair Dunlop, who was instrumental in forging the collaboration, commented: ‘Up until now, scientists could only talk to their own networks. This software makes it possible for them to talk across various networks and increase the range of computational power available to e-scientists.’
Another project using the interoperable grid software is the Virtual Physiological Human (VPH), which seeks to create a comprehensive computer model of the human body. Once established, the methodological and technological framework will enable the investigation of the human body as a single complex system. The Wellcome Trust Heart Physiome Project, for example, is a collaboration between the Universities of Oxford in England and Auckland in New Zealand to develop a multi-scale modelling framework for the heart, which can be used for addressing scientific and clinical questions.
Other applications for interoperable grid software include investigations into clean-power generation. E-scientists are studying plasma physics as a possible way of producing electricity in fusion-power plants. Complex computer modelling is required that demands high levels of computer processing. EUFORIA (EU Fusion For ITER Applications) is a consortium of European partners, including Edinburgh University. EUFORIA enhanced modelling capabilities for fusion scientists by deploying OMII-Europe components to enable them to simulate the full-fusion process seamlessly using resources in EGEE and DEISA.
OMII-Europe provides worldwide leadership on the integration of major grid computing platforms. Its implementation of emerging open standards in its open-source interoperable software directly benefits the e-science community and has huge implications for scientific collaboration in the future. As Steve Brewer says: ‘There is an ongoing need for more work in this area, and projects of this sort are the way to bring experts on board. Gathering together the different technical and research communities was very challenging, but has exemplified the process for broadening the horizons for
e-scientists in Europe and beyond.’
Sara Goodwins is a freelance writer specialising in education and business.
Discover other Computer Science & IT research papers, including a 2006 about the Dare To Be Digital Computer Games Competition and 2007 reseacrh about SimCity For Real.
