ISC HPC Blog
Mediterranean Mega computing
In 2008 I moved from Australia to the equally sunny, but significantly smaller, island of Cyprus to become one of the scientific coordinators at the recently founded Cyprus Institute. More particularly, I was a member of the Computation-based Science and Technology Research Centre (which thankfully was abbreviated to CaSToRC soon afterwards). One of my main roles there was to work on the LinkSCEEM project.
The project was being supported by the European Commission FP7 programme to develop research infrastructure and the basic idea was to attempt to establish a culture of simulation-based scientific research in the Eastern Mediterranean. The Cyprus Institute wanted to host a large HPC center but they didn’t have the population in Cyprus to sustain it. However, if you draw a 200km ring around Cyprus though you encircle some 200 million people in places like Egypt, Jordan, Israel, Turkey, Lebanon and Greece. Some of these have great technical advantages and expertise (like Greece’s excellent network connectivity or Israel’s booming IT) but most lacked any real HPC resources: at the time there were just 3 entries in the entire region in the Top500 list. If they did have HPC resources in the past, it tended to have been well funded initially but sustainability was hard once the first machine started to age. Cyprus did not want to see this repeat itself and the idea was to tap the scientific human resources of the region to help ensure the sustainability of a large facility: if scientists were using the facilities and producing high quality science, ideas and students then the facilities are demonstrably worth maintaining.
That first 2 years was then spent visiting the region, talking to scientists, having training events and generally trying to figure out how to create critical mass. People were mostly enthusiastic and sometimes skeptical, which is understandable if you have seen facilities close in the past and not be replaced. The students and early career researchers we met were hungry but the experience or resources to train them were not readily available. A wide spectrum of knowledge existed, some people had used huge resources in the US and Europe, others had never done anything outside their desktop. This was exacerbated by the fact that experienced people taught in their own discipline only which hindered the spreading of knowledge.
The many thorny details also became apparent. Lack of local expertise was one thing, but network connectivity reared its head as the elephant in the corner. I’d been spoiled in Ireland and Australia in the past, which had invested heavily in networking since they saw it as an invaluable tool for IT research, development, and industry. The Eastern Mediterranean was completely different, there were universities that could not provide email accounts to students because their internet connection would be overloaded by the resulting spam traffic. The cost differences when compared to mainland Europe were astonishing, orders of magnitude more expensive. For countries with much less financial resources, this was crippling for their universities. Luckily things have recently started to change in that respect. Increased competition has lowered prices and improving network connectivity has become a central component of the second phase of the LinkSCEEM project which began in 2010.
Since 2010, I’ve been working with Juelich Supercomputing Centre which, along with the National Centre of Supercomputing Applications in the US, act as helpful big brothers to the development efforts in the region. This year the 40 teraflop Cy-Tera hybrid cluster was installed in Cyprus and the programs have already begun for access to this and the 2 other available systems available in Egypt at Bibliotheca Alexandrina and the National Authority for Remote Sensing and Space Sciences. The concept has also extended beyond cycle-based HPC and also includes high performance data systems and image capture and visualization technologies.
The project is working on a program of training and integration events over the next few years to bring people to the resources and also create research collaborations that cross borders. Some prototype research areas have been especially encouraged to give this effort a kick-start. These areas are chosen because they are considered to have regional significance, with research communities that span many countries in the Eastern Mediterranean: climate studies are an important guide for the future of the water-parched countries of the region, synchrotron radiation research complements the SESAME facility being built in Jordan and digital cultural heritage allows the recording, collection and contextualization of artifacts from the “cradle of civilization” that are now distributed all over the world.
Seeing the original concept begin to bear fruit makes this an exciting, and frequently challenging, time.
About the author:
Dr. Alan O'Cais has been active in the field of computational research and high performance computing since receiving his bachelor’s degree in Theoretical Physics from Trinity College Dublin in 2001. He received a Master’s Degree in High Performance Computing in 2002 and a PhD in
Lattice Quantum Chromodynamics in 2005. In 2008 he joined the Cyprus Institute as Scientific
Coordinator and in 2010 moved to Juelich Supercomputing Centre where he works within the Application Support division. His work is primarily focused within the LinkSCEEM-2 project helping to develop a virtual research community in computational science in the Eastern Mediterranean region.