When Eumetsat was planning for a near real-time data processing function for its Meteosat Third generation programme (MTG) it became apparent that this would require an unprecedented level of computing power. Its on-demand re-processing of batch data pushed compute requirements to a speed up to 30 times faster than real time with much higher data volumes than seen in the existing systems.
We have led the feasibility study and the design of prototypes to evaluate grid, cloud and supercomputing architectures to determine the most suitable option and right technologies for the MTG programme to thus enable a long-term reduction of costs.