Project: ePortal (2000-2001)
EPCC’s ePortal Project focused on the Grid concept of “portal computing”. In this model the user interface, rather than user source code, is portable across Grid computing resources. The difficult job of porting application codes between platforms is undertaken, as it should be, by the application provider, not the user.
In the ePortal model, the system as a whole is restricted to running known and trusted applications. Providers of services for computational scientists - regional HPC centres like EPCC, for instance - would register with the ePortal whatever applications they wished to offer. Because these applications are compiled and configured entirely by the service provider, they can be made secure so there is no need to provide each user with their own login account - all jobs run under the account of the application service provider. The end user does not have to know what system their code is running on, only which service, and perhaps which service provider, they wish to use. Some service providers may load balance their workload across several systems; others may correspond to a particular machine.
The EPCC ePortal offered a single view of a range of available software packages for computational chemistry, fluid dynamics, crash simulation etc. A scientist was able to select a service to run their particular jobs using their favourite package, but in a transparent way that freed them from the details of the computing and allowed them to concentrate on the science. No matter whether their jobs would run on a Cray T3E, an SGI Origin system or a cluster of Sun HPC servers, the interface to, say, their chosen chemistry code was the same.