Computing in a grid or a cloud
In the past year or two, cloud computing (in which computing services are all extracted from the "cloud" of the Internet) has garnered a lot of attention in the press. We talked about its application to high-energy physics here back in May.
Meanwhile, high-energy physics has been working hard to develop grid computing, and one fruit of that development is the Worldwide Large Hadron Collider Computing Grid, which just launched on October 3, 2008.
With two relatively new models of computing floating about, many commentators have been discussing the pros and cons of each, as if it were a face-off with only one likely to be the victor in a fight for dominance, and with cloud computing being a newer concept, some even argue that cloud computing will supersede grid computing. However, an analysis by Ignacio Martín Llorente of the Distributed Systems Architecture Research Group in Madrid, argues that the two technologies are different and complementary. (Thanks to International Science Grid This Week for the link.)
He says that grids are all about interoperability, allowing a lot of different resources to be federated--that is, to come together to work effectively. A cloud, on the other hand, is principally a way to virtualize services (such as computation power or data storage), or take them away from your local place of work and put them wherever they most conveniently sit but still available to you at all times.
I'd recommend Llorente's post to help understand this better, but we also have a few relevant resources here at symmetry about grid computing:
The Grid in 60 seconds
Meet the Grid (An introductory feature)
What is the Grid? (A conversation between some leading particle physics computer scientists)
Sciences on the Grid (Case studies of science being performed on existing grids)