Research Activity

2. High performance and  grid computing

What is grid computing?

Several definitions exist but they converge toward a same idea. Indeed, grid computing is a shared set of varied and geographically distributed resources across multiple administrative domains and owned by different organizations.

Inspired by the electrical power grid, grid computing aims to offer to the user a reliable and transparent access to the grid resources regardless their geographical sites. 

Why grid computing?

The huge quantity of computing resources dispersed everywhere in the world leads to the important idea of better exploiting these resources by using them when they are idle. 

Regrouping these resources provides an important parallel computing capacity which permits to replace supercalculators and parallel machines that are very expensive. 

Considering the big number of resources federated by the grid, the fact that some of them are inaccessible doesn't damage its reliability and doesn't touch to the continuity of its services.

Research topics

Grid computing promises to radically change the way we discern and use high performance computing: the challenge consists on exploiting worldwide computational resources in order to create a source of a big computing power able to support complex applications.

Grid computing represents a privileged research axis in UTIC, since this new technology is a promising tool for the federation of computer resources.

Several works are launched in UTIC and aim to touch to this new technology closely. Indeed, UTIC is interested to create several resources management tools for the grid by conceiving new approaches and algorithms of scheduling for the grid.

Considering the huge quantity of data stocked and treated by the applications of control and management of grid resources, it is important to apply data mining algorithms for the analysis of data warehouses in order to find the needed information to better exploit resources.

Another prospective idea is the application of economic models in order to give an economic behaviour to the use of the grid resources.

Another interest concerns the parallel sorting algorithms; indeed, we suggest that these approaches can be useful to better manage grid resources and to optimize some applications on the grid.

research activity