Next Generation Grid: Learn from the Past, Look to the Future

Abstract

The first part of this talk will be focused on the Grid Evolution. In fact, in order to discuss “what is a Next Generation Grid”, it is important to determine “with respect to what”. Distinct phases in the evolution of Grids are observable. At the beginning of the 90’s, in order to tackle huge scientific problems, in several important research centers tests were conducted on the cooperative use of geographically distributed resources, conceived as a single powerful computer. In 1992, Charlie Catlett and Larry Smarr coined the term “Metacomputing” to describe this innovative computational approach [1].

The term Grid Computing was introduced by Foster and Kesselman a few years later, and in the meanwhile several other words were used to describe this new computational approach, such as Heterogeneous Computing, Networked Virtual Supercomputing, Heterogeneous Supercomputing, Seamless Computing, etc.,Metacomputing could be considered as the 1st generation of Grid Computing, some kind of “proto-Grid”.

The Second Grid Computing generation starts around 2001, when Foster at al. proposed Grid Computing as an important new field, distinguished from conventional distributed computing by its focus on large-scale resource sharing, innovative applications, and, in some cases, high-performance orientation [2].

With the advent of multiple different Grid technologies the creativity of the research community was further stimulated, and several Grid projects were proposed worldwide. But soon a new question about how to guarantee interoperability among Grids was raised. In fact, the Grid Community, mainly created around the Global Grid Forum (GGF) [3], perceived the real risk that the far-reaching vision offered by Grid Computing could be obscured by the lack of interoperability standards among the current Grid technologies.

The marriage of the Web Services technologies [4] with the Second Generation Grid technology led to the valuable GGF Open Grid Services Architec-ture (OGSA) [5], and to the creation of the Grid Service concept and specifica-tion (Open Grid Service Infrastructure – OGSI). OGSA can be considered the milestone architecture to build Third Generation Grids.

The second part of this talk aims to present the outcome of a group of independent experts convened by the European Commission with the objective to identify potential European Research priorities for Next Generation Grid(s) in 2005 – 2010 [6]. The Next Generation Grid Properties (“The NGG Wish List”) will be presented. The current Grid implementations do not individually possess all of the properties reported in the NGG document. However, future Grids not possessing them are unlikely to be of significant use and, therefore, inadequate from both research and commercial perspectives. In order to real-ise the NGG vision much research is needed.

During the last few years, several new terms such as Global Computing, Ubiquitous Computing, Utility Computing, Pervasive Computing, On-demand Computing, Autonomic Computing, Ambient Intelligence [7], etc., have been coined. In some cases, these terms describe very similar computational approaches. Consequently, some people are raising the following questions: Are these computational approaches facets of the same medal? The last part of this talk will explore the relationship of these approaches with Grid.