e-infrastructure projects around the world are working to integrate access to supercomputers, grids and clouds. Although there are benefits of having all the processors needed for a complex calculation in one place – such as removing the lag time or ‘latency’ experienced when the calculation is done over grid – network speeds are increasing to the point where such issues are becoming negligible.
Some projects working on usability applications aim to deliver services intelligently, whether through HPC or grid, based on the needs of the user. The application itself can decide whether the question the scientists are trying to answer needs HPC or is better suited to grid, for instance.
There are also some services, both commercial and non-commercial, that are bringing HPC to the cloud. In this scenario, organizations can hire the use of a virtual supercomputer, based on the large-scale clusters in so-called processor farms.