The RealityGrid is an EPSRC funded pilot project to examine how the condensed matter, materials and biological sciences communities can make more effective use of distributed scientific computing and visualisation resources within their future research activities. Existing terascale computing environments are able to generate data at a rate several orders of magnitude beyond our ability to extract the knowledge produced during the simulation. Simulations take several days to run but the data analysis still takes months! The goal of the RealityGrid project is to generate a new computational science analysis pipeline that "moves the bottleneck out of the hardware and back into the human mind."

ICENI - A Grid Enabled Component Middleware

The London e-Science Centre is supporting the 'deep track' activities within the RealityGrid through its next generation Grid Middleware - the Imperial College e-Science Networked Infrastructure (ICENI) - which uses a component programming methodology to define how individual components may be composed to build a complete application. Through meta-data associated with each component we can make automatic deployment decisions to optimise the placement of the resulting application on distributed Grid resources. We can also use this meta-data information to monitor or alter the data stored within a component.

The ICENI framework exposes the functionality within an application as software components. These are composed within a visual programming environment to build a complete application that is then passed to a scheduler service. The scheduler matches the user-defined application to the component implementations that are currently available on the Grid resources. The 'best' resources (a user defined criteria) are then selected for application execution. The components within a running application are exposed as services so that they may be manipulated, if policy permits, during execution.

User 1 is only able to read information from the application while User 2 is able to both read and write to the application and therefore has full control over its behaviour. The user who initiates the application may also define a policy specifying who will have the required permissions to manipulate its execution.

Collaborative Scientific Visualisation & Steering within ICENI

Collaborative scientific visualisation and computational steering are key requirements within the RealityGrid project that need to be supported by ICENI - the underlying Grid Middleware. The ICENI component framework offers us the benefits of encapsulation and code reuse that allows us to carry out high-level composition of complex applications and to describe how other entities may interact with it. The ability to describe the interaction characteristics between components allows us to optimise deployment. In the context of scientific visualisation, a post-processing stage may be co-located with the application before the data is transferred to a computational resource local to the user for rendering before the pixels are moved to a local display device.

Such a distributed application may also be steered. In the following example collaborators at different locations view the results from a running application.

Collaborators at different locations view the results from a running application.