Grid: More bytes for science

What is there in common between the fight against avian flu, the development of drugs against malaria, the quest to understand the first instants of the Universe and research on climate change? All these topics require a huge amount of computer power and data storage capacity that can be satisfied with the Grid.

In the last week of September, over 500 scientists and software engineers from Europe and around the world met in Geneva for a conference devoted to the Enabling Grids for E-sciencE (EGEE) project, the global Grid computing infrastructure project led by CERN, which is changing the way science is done. One example that the experts assembled in Geneva discussed is how Grid computing is helping in the battle against avian flu and other diseases.

During the month of April, the Grid was used in the fight against the lethal H5N1 avian flu virus. Thanks to the Grid infrastructure of the EGEE project, six laboratories in Europe and Asia analysed 300 000 chemical compounds in search of potential drugs for treating the disease. This research, which involved 2000 computers linked together around the globe, in part thanks to software developed at CERN, helped to identify the most likely compounds able to inhibit the enzyme N1 that is part of the virus. In one month, the collaboration achieved the equivalent of 100 years processing on a single computer.  

As well as avian flu, the EGEE infrastructure recently contributed to simulations of over 40 million potential drug molecules against malaria. Grid computing can help drug discovery by speeding up the computer-based screening process of drug candidates and  thus reducing the cost of new drug development to treat diseases such as malaria.  In the coming months, an even more ambitious challenge - the equivalent of 500  years of processing on a single computer - is planned to test drug molecules proposed by research groups from Italy, the UK, Venezuela, South Africa and Thailand against a range of so-called neglected diseases, which threaten livelihoods in developing countries.

The Grid is based on the same idea as the Web, which was invented at CERN in the beginning of the 90s: sharing resources between geographically distributed computers. But whereas the Web simply shares information on the computers, the Grid also shares computing power and storage capacity. This means that scientists can log on to the Grid from their PC, and the work they need to be done will be carried out by many machines across the planet. This allows scientists to carry out very complex calculations quickly and simply.

CERN is interested in Grid computing because it has to process a lot of data. This need will shortly get even bigger when CERN's new particle accelerator, the Large Hadron Collider (LHC) begins work in 2008. This accelerator, which will be the world’s largest scientific instrument, is predicted to produce 15 petabytes a year or nearly 1% of all the digital information produced annually on the planet. This is equivalent to 15 million gigabytes, which could be stored in a stack of CDs  more than 20 km high. The Grid provides a solution to how to store and process such huge amounts of data on a global scale, using dedicated optical fibre links to sustain data transfer at rates in excess of one gigabyte/second – fast enough to download an entire DVD in a few seconds.  

CERN leads two major Grid projects, the first is the LHC Computing Grid (LCG), which is dedicated to providing the processing power and storage capacity necessary for the LHC. The second, which is co-funded by the European Commission, is called Enabling Grids for E-sciencE (EGEE). The EGEE Grid is used by a wide range of research communities, from biomedical science to astrophysics, and including industrial applications, all sharing a common infrastructure. At present this Grid has nearly 200 sites around the globe, connecting the power of over 20 000 computers into a seamless whole that runs up to 30 000 jobs per day. It is expected to grow considerably during the next two years.

In addition to these projects, the CERN openlab industrial partnership provides a framework for CERN and leading IT companies to collaborate on long-term software and hardware developments for the Grid.

Grid map courtesy of Imperial College Real Time Monitor.

September 2006