The history of Grid
The term Grid was first introduced by Ian Foster and Carl Kesselman as an analogy with the electrical power grid indicating that computing power will be accessed in such an easy and transparent way as electric power is now acquired through a plug in the wall.
The term was not used before the late nineties, but before that several technologies such as Beowulf clusters and CPU-scavenging software had greatly influenced the evolution of key Grid concepts.
Beowulf cluster was the first network of inexpensive personal computers put together in order to work as a single super-computer able to perform tightly coupled parallel HPC computations. The cluster was originally designed and employed by NASA in 1995 but Beowulf clusters are still widely used by scientific communities worldwide.
CPU-scavenging introduced the idea of splitting computationally demanding jobs to small sub-jobs that can run in parallel and distributing them to a network of computers to execute them at times when they would otherwise be idle. A major project in this area was seti@home (1997) which used the CPU power from a network of volunteers with internet enabled computers around the world in order to look for Extra-Terrestrial Intelligence signals. Although the project failed to detect any signs of intelligent life outside the earth, it did succeed in displaying the efficiency of distributed grid computing.
The first project that differentiated grids from other distributed networks by moving the focus from geographical distribution to software integration and management was I-WAY (Information Wide-Area Year). The experiment, led by Ian Foster, put together a network of 17 super computer centres, 5 virtual reality research sites, and over 60 applications groups in order to set up a widely distributed test bed for advanced applications.
Most importantly, the experiences gained and software developed as a result of the I-WAY project was used as input for building Globus, a middleware infrastructure which was first introduced in 1996 allowing applications to be deployed on a network of distributed and heterogeneous resources appearing to the application as a single virtual machine. A central element of Globus is the Globus Toolkit which defines the basic services and capabilities required to construct a computational Grid and which still constitutes the base technology for the largest part of Grid projects today.
In 2002, in order to address the issue of standardisation in grids, Ian Foster, Carl Kesselman, Jeffrey M. Nick and Steven Tuecke in their paper “The physiology of the Grid” proposed an architecture defining a set of core capabilities and behaviours that address key concerns in Grid systems. The Open Grid Services Architecture (OGSA), developed by Global Grid Forum working groups, represents an evolution towards a Grid system architecture based on web services concepts and technologies.
In parallel with the evolution of Grid computing, much work was being done in the web services community to define specifications related to the creation of secure, interoperable web services. In an effort to align grid services technologies with web service standards, in 2004 OASIS introduced the Web Services Resource Framework (WSRF), a set of WS specifications describing how to implement OGSA capabilities using web services. WSRF constitutes an important milestone in the evolution of Grid computing and a major step towards making grid computing a key technology in the next generation internet.
