Grid Overview
"The Grid", in this context, really means a collection of infrastructure based on standards that will allow researchers to use remote, probably large, compute and data engines without having to know specific details about them. With more and more research depending on heavy duty ICT capabilities, it's unreasonable to expect researchers to develop and maintain the necessary skills to use all these systems.
Instead, the grid aims to present the researcher with a reasonably uniform interface, one that's well documented, easy to learn and hopefully static. When HPC (high performance compute) or Data systems change, it's up to the operators of that equipment to make sure they connect the interface that the end use expects to the new system. As they say, "it just works" !
The grid consists of a number of components. Traditionally some of these components have been quite complicated themselves, however, ARCS has done considerable work to ensure that there are easy workarounds while still leaving the raw power of the grid available to those super users who want just that.
| Authentication | If you are to use resources, the resource provider needs to know that you are entitled to use them. The Grid has traditionally used X509 certificates. These are not easy for the end user to mange so ARCS has built Grix to solve this problem. We believe that the Federally funded AAF project will make life even easier soon! |
| Authorisation | Again, these resources are expensive and often require specific permission to use. ARCS currently uses VOMRS to manage authorisation. Alternatively, its possible for you to map your grid credentials directly to your pre existing logon account at a particular site. |
| Data Movement | ARCS recognises that end users need to move data around for a number of reasons, perhaps as input or output to or from a compute job, perhaps to distribute it between a team, perhaps to archive, annotate or distribute to the general public. |
| Compute Jobs | While the grid was initially seen as a gateway to HPC, it's more and more seen as a means to provide easy job launch for those jobs a bit beyond the desktop. End users often expect easy "point and click" Graphical User Interfaces (GUI) and ARCS has a number of tools and the skills to produce just that. Grisu and Nimrod are good examples. |
| Workflow | One of the strengths of the Grid is its ability to automate complicated work flows. Because the interfaces to the grid remain constant, it's reasonable to invest effort in a workflow engine knowing that it will be capable of using a large number of "back end" systems and won't break when someone decides to upgrade their HPC engines. |