The idea of “grid computing” was put forward for the first time by the computer scientists Ian Foster and Carl Kesselman in 1998. They recognized that in the natural sciences, both the number of calculations and the raw amount of data are rapidly starting to exceed the capabilities of what can be reasonably be done at single institutions. One possible solution is to connect supercomputers and compute clusters across institutions and across nations.

Foster and Kesselman’s vision calls for a compute grid that is analogous to the power grid. In the future, users should be able to obtain compute capacity on-demand, and as easily as electricity today. To cook myself a coffee, I don’t need to know which power plant is currently providing my electricity, and what type of generators it is using. Just like that, grid calculations should work: Computers with free capacity are automatically located, and the requested application is executed there on the necessary data. None of this should need the user’s intervention, and results are automatically returned.

One of the most important driving forces in grid computing is the CERN in Geneva. High-energy physicists were the first to embrace grid computing and to adapt it to their needs. Their efforts have resulted in the first grid implementations and in a clear orientation towards open standards and open-source software.

In the meantime, other scientific disciplines, in particular life sciences, have discovered grid computing for their needs. A number of initiatives, such as the SwissBioGrid or HealthGrid, aim to find and adapt suitable applications from their domain to grid computing, but also to adapt the grid infrastructure to their particular needs and research questions. The aim of SwissBioGrid is to provide its partners a platform that allows the computation, modeling and simulation of complex systems and questions in the domains of computational biology and computational chemistry.

Motivated by the dengue project, I have been involved with SwissBioGrid from its founding moments on, and have worked on the development and implementation of its prototype middleware, ProtoGRID.

No Comments »