Grid on Demand

Grid computing has traditionally been the dedicated computing platform for compute intensive scientific (e-Science) applications. Grid is developed by the Grid community to share resources of the existing dedicated compute clusters and to enable large-scale e-Science experiments. Because of Grid those applications and experiments potentially can leverage the sum of the compute capacity of all participants. Still, orchestrating fair sharing among the participants is a complicated affair. Users of the Grid might sometimes need more than they have immediately available.

Slides of the Presentation (30 july 2010) of Grid on Demand: Link to PDF

Cloud computing is emerging and offers compute resources as a service. Large scale e-Science applications with a sudden demand for compute power (urgent computing) can benefit from Cloud's immediate availability of resources and the predictability of the type of resource leased. The ability to quickly, but temporarily, utilize massive amounts of compute power to have the result as quickly as possible, would under such circumstances be very desirable. An on demand Grid could provide this ability. Combining the properties of both Grid and Cloud for the purpose of supporting current or newly developed e-Science applications with a sudden demand for compute power, the report investigates the following research question: “Can Grid computing be offered as a Cloud service”.

The Grid on Demand Research Report: Link to PDF

Grid on Demand version 1.22 is available through the Public Amazon Machine Images listed below:

region 32-bits 64-bits
us-west-1 ami-97da8bd2 ami-41dc8d04
ap-southeast-1 ami-0dc3bc5f ami-0fc3bc5d
eu-west-1 ami-9bd8f2ef ami-85d8f2f1