Home » Technology » Scientific Computing » A Cycle is a Terrible Thing to Waste

A Cycle is a Terrible Thing to Waste

© Adam Hughes

In the world of computing in the late (very late!) 1990's, one of the overriding themes is how to best utilize the millions of computer cycles that are burned around the globe each day. Indeed, the cyber landscape is rife with "live" machines that do little other than send e-mail or browse the web as their main functions, but whose CPU's keep ticking all the while. With the increasing demand for computational workhorses brought about from the simulation researchers in the scientific community as well as others, this vast resource is becoming a very attractive target. This same situation is at the heart of the burgeoning "cluster" movement, in which a group of machines are linked to form a sort of scaled-down supercomputer. While clusters have been successfully established at a number of institutions, some people are thinking bigger. The natural successor to the cluster idea is coming to fruition as computer scientists work to create technologies that will enable the use of geographically-distributed computer resources. One of these schemes, Globus, is enjoying particular popularity among scientific groups and promises to forge into untamed cyberspace to wrange compute cycles for those who could use them.

Globus is a project being developed at Argonne National Laboratory in Illinois to take advantage of precisely the situation outlined above. The objective of Globus is to develop the basic technology that will enable an application code to take advantage of what it calls "computational grids", which are sets of geographically-distributed machines available from the user's desktop. The Globus project consists of four main components : Research, Tools, Testbeds, and Applications. Each of these areas is described in detail at the Globus web site (http://www.globus.org), but they each merit a brief discussion here to begin to better understand this concept.

Like any project on the cutting edge, Globus is inherently a research proposition. The two major areas of research relevant to gridding technology are resource management and security. In order to effectively utilize the computational resources around the world, Globus must be able to identify which sources are available and suitable for handling a certain task, and it must provide for the security of the information to be transmitted between the workstation client and the global server. These are not trivial tasks in a simple cluster environment, and the complications are obviously of a much grander scale when discussing global computing.

In addition to developing the grid technology necessary to make distributed computing possible, Globus also strives to provide the computational tools and services that will allow others to build such grids. The Globus "toolbag" will be a group of

The copyright of the article A Cycle is a Terrible Thing to Waste in Scientific Computing is owned by . Permission to republish A Cycle is a Terrible Thing to Waste in print or online must be granted by the author in writing.