A classification of emerging and traditional grid systems

Abstract

The grid has evolved in numerous distinct phases. It started in the early ’90s as a model of metacomputing in which supercomputers share resources; subsequently, researchers added the ability to share data. This is usually referred to as the first-generation grid. By the late ’90s, researchers had outlined the framework for second-generation grids, characterized by their use of grid middleware systems to “glue” different grid technologies together. Third-generation grids originated in the early millennium when Web technology was combined with second-generation grids. As a result, the invisible grid, in which grid complexity is fully hidden through resource virtualization, started receiving attention. Subsequently, grid researchers identified the requirement for semantically rich knowledge grids, in which middleware technologies are more intelligent and autonomic. Recently, the necessity for grids to support and extend the ambient intelligence vision has emerged. In AmI, humans are surrounded by computing technologies that are unobtrusively embedded in their surroundings.

However, third-generation grids’ current architecture doesn’t meet the requirements of next-generation grids (NGG) and service-oriented knowledge utility (SOKU).4 A few years ago, a group of independent experts, arranged by the European Commission, identified these shortcomings as a way to identify potential European grid research priorities for 2010 and beyond. The experts envision grid systems’ information, knowledge, and processing capabilities as a set of utility services.3 Consequently, new grid systems are emerging to materialize these visions. Here, we review emerging grids and classify them to motivate further research and help establish a solid foundation in this rapidly evolving area.