An invitation to participate in a green computing project was issued by Bill St. Arnaud of CANARIE during the TNC 2009 plenary session yesterday, Tuesday 9 May. He gave an overview of evidence for global climate change and the reductions in carbon dioxide emissions that have been recommended in order to slow the rate of change enough so that we have time to adapt. “This will affect research networking along with every other aspect of life and society,” he said.
The high energy demand of ICT departments in research universities makes them vulnerable to impending policies and regulations that may impose penalties on heavy carbon emitters. But Bill St. Arnaud said there are opportunities in this situation too. The trade in carbon offsets is expected to become a 645 billion dollar industry in the near future and the research and education community could benefit from early participation in this market, if we are well prepared.
Moving data centres close to their own sources of relatively cheap renewable energy would facilitate a zero carbon strategy. Wind and solar power would be the most suitable sources, said St. Arnaud. When the wind stops blowing or the sun goes down, whole data sets or computing jobs could be transferred to computing clouds with available power so that work would continue. The challenge would be to design network architectures and business models that could ensure reliable service delivery and avoid greatly increased energy consumption by the network.
National research and education networks (NRENs) will have to help find solutions, St. Arnaud said. He invited TNC participants to get involved in a new green cyber-infrastructure project to build a zero-carbon NREN and to show how to capture revenue from this. “We can be important leaders and show how networks can play an important role in addressing the biggest challenge facing our planet,” he concluded.
Clouds for e-science
In his plenary presentation, Paul Watson from the University of Newcastle focused on the advantages cloud computing can bring to science. Drawing on lessons from web-based companies that cope with peaks and dips in demand for their services by using servers in computer clouds, rather than their own hardware, he described a UK project which uses clouds to store and analyse data, and to share data and information with varying degrees of security.
He demonstrated the system to TNC participants, explaining that “clouds reduce the time from having an idea to realising it” by increasing the number of CPUs at the touch of a button, rather than with a lengthy planning and provisioning of hardware. “Clouds can revoutionise e-science,” said Paul Watson, by helping to reduce complexity. People are now thinking about federating private clouds, he said, so you could build service agreements with public clouds. “This will become important in the future because of the need to deal with renewable energy and other issues”.
In an afternoon session on climate change and disaster forecasting, two talks discussed how computer modelling can be combined with information from the real world to predict the likely impact of catastrophic events. Networks and geographical information systems can be used to incorporate information about population densities and building types to provide decision makers with rapid estimates of how severely people and infrastructures are likely to be affected. Mathematical modelling can also be used to understand physical processes, for example how landslides can generate tsunamis. A theoretical model has been combined with information about the actual seabed of the Mediterranean to assess the risks from such landslides in future.
The risks of revealing personal information were discussed in a session called, “Implications”, which featured an attempt to bribe attendees into giving away their names, passwords and other information. John Paschoud (LSE) told participants that a similar exercise to benchmark security among students showed that large percentages of them were willing to reveal various details in exchange for chocolate. Security messages on pub beermats were remembered by students to some extent so, he suggested, this kind of approach to raising awareness may be “more effective than yet another policy”.
In a popular session called, “All Change”, Hans Doebbeling (DANTE) presented an overview of the new GN3 project. He emphasised that, whereas GN2 had concentrated on building the GÉANT network, GN3 focuses on the user and providing high performance and services. Afrodite Sevasti (GRNET) addressed how the research, development and roll-out of multi-domain and end user services will be achieved. In the same session, Steve Cotter presented the new hybrid network architecture of ESnet (US Department of Energy's Energy Sciences Network) and how it will meet future needs of virtual collaborators and science. Work was completed in November, but the company is busy with a 70 million dollar grant from ARRA (American Recovery and Reinvestment Act), to stimulate and accelerate research into networking technologies.
Two sessions were dedicated to the FEDERICA project, which is closing the gap between middleware services and network infrastructure, and exploring the limits of virtualisation. How to transform the network infrastructure into a service accessible to users was the subject of one presentation, followed by the challenges of measuring in a virtualised environment. The relevance of the future Internet in the European Commission’s plans for the coming years was also explored.
Archived video streams of all TNC presentations are available from http://tnc2009.terena.org/media/archive.php.
Most of the slides and some of the papers can be downloaded by clicking on the relevant session in the programme, at http://tnc2009.terena.org/schedule/.