1.1 Introduction

Computational grids find their origins and background in the field of HPC, with the preeminent goal of linking supercomputing sites and optimally exploiting CPU time available through a wide area multidomain networking connection. It is a common situation, indeed, that on a certain node, at a given instant, a strong computational effort must be sustained, while huge CPU-time amounts are left idle on remote sites. The existence of a pervasive intelligence could monitor the status of each processor, assigning CPU power where needed in a right-sized fashion, thus reducing as much as possible idle CPU times and allowing controlled access to large-scale CPU facilities. The role played by the evoked pervasive intelligence is starred by what we call now a computational grid.

After (or, more realistically, while) achieving the goal of facilitating and improving HPC, grids have naturally evolved: the same idea of controlled and optimized management of available distributed CPU power has been extended to the more general concept of resource management. As you may need CPU time at a certain moment, you may also want to access a remote database, store large amounts of data on remote storage systems, and access software services and electronic instruments. Grids are in charge of enabling users to do this.

In such a new and variegated context, which evolves along with the impressive Internet revolution and its Web facilities, grids have grown far beyond their early conception, representing the natural melting pot of distributed systems, networking, security, parallel...