Last week a new super-data-cluster was installed by SGI® and Panasas® in the STFC e-Science data centre at the Rutherford Appleton Laboratory which will be used by researchers studying climate change and processing satellite data. The JASMIN+CEMS cluster combines two machines into a single £4.5 million, 10 ton hardware system (75% JASMIN, 25% CEMS).
STFC e-science host the JASMIN machine in the RAL data centre on behalf of RAL Space and the National Centre for Atmospheric Science (NCAS, a research centre of the Natural Environment Research Council, NERC). The CEMS machine is hosted on behalf of the International Space Innovation Centre (ISIC) Facility for Climate and Environmental Monitoring from Space (CEMS) on the Harwell campus, itself formed as a collaboration between industry, academia and government in order to actively promote ways in which other market sectors can use space-derived data and technologies to develop new products and services to enhance their businesses. The CEMS service has already enabled four consortia to win funding for ‘fast track’ projects from the recent ‘Space for Growth’ competition within the National Space Technology programme.
JASMIN+CEMS will replace the existing computing resources and NFS storage for 1.2 Petabytes of data in the Centre for Environmental Data Archival (CEDA) at RAL Space, allowing data to be managed with fewer people and provide faster access to data, as part of a single efficient scientific data centre at RAL. It will also support efficient data analysis by the UK and European climate and earth system science communities, basic processing of earth observation data, and new ways of supporting flexible access to high volume and complex climate and earth observation data.
The JASMIN+CEMS cluster includes 4.6 Petabytes of usable fast access Panasas® parallel file storage. The important aspects of the data storage design are the 1 Tb/s aggregate bandwidth from data to processors which supports the processing of very large data volumes, and the lower total cost of ownership than competing solutions due to less need for manual intervention by operators to manage and expand the system. The 1133 data blades constitute the second largest configuration that Panasas® have provided to a single installation. The largest configuration was of 1166 blades for the Roadrunner supercomputer at the Los Alamos National Laboratory in New Mexico, USA, which headed the list of the world's top 500 computers in 2008. However, since storage technology has moved on in the last four years, JASMIN+CEMS has a larger total storage capacity than Roadrunner had then.
For processing, the JASMIN+CEMS cluster also includes 27*12 core and 1*48 core processors. The processors and data storage are configured for a combination of data serving and number crunching which can be switched from one to the other depending on demand.
The funding for both JASMIN+CEMS comes to STFC from the UK government investment of £145 million in e-infrastructure announced in 03 October 2011 by the Department for Business, Innovation and Skills – but JASMIN’s funding comes via NERC, and CEMS via ISIC and the UK Space Agency.