About

Over the last ten years the European countries and the EC have made significant investments in e-Infrastructures for scientific computing, notably High Throughput Computing (HTC) and High Performance Computing (HPC) services. Their contribution in pushing the European research towards a leading position and addressing global challenges has been validated by a growing number of research initiatives, most notably the ESFRI projects. Sustainability of such services is crucial, as the research supported by them is more and more crucial for European competitiveness – and sustainability can be planned better if the costs are known.

The emerging commercial offerings (Cloud-based HTC and HPC solutions) pose additional challenges and opportunities for sustainability. They represent new ways of explicit monetisation of computing and open the way to new business models (e.g. service fees or of lease industrial resources). The upcoming next framework programme and broader EU policies depend on more careful impact assessment, which requires accurate cost estimates. Similarly energy saving and green computing will be increasingly relevant, and project results will also allow estimating potential benefits in these areas. It is thus vital to understand the cost of the dedicated computing related e-Infrastructures, as expressed by EGI and PRACE initiatives. The calculation of such costs is not trivial; keeping detailed accounting data for the evolving mix of capital (hardware) and operational (personnel, energy) expenditures is only one of the challenges. The scope of the e-FISCAL project is to analyse such costs, facilitated by national entities (NGIs and HPC centres), compare them with equivalent commercial leased or on-demand offerings and provide an evaluation report. The execution of the survey builds on a small-scale survey conducted by the e-IRGSP2 project that proved the complexity of the issue and provided tacit knowledge valuable for the e-FISCAL project.

Objectives:

The overall goal of the project is to contribute to a more accurate understanding of the costs of HTC and HPC services, both at a national and European level. The project aims to achieve this goal through the following activities:

1. Study the dedicated HTC and HPC e-Infrastructure costs by surveying European National Grid Initiatives (NGIs) and national (or pan-European) HPC centres and analysing cost factors to better understand what e‑Infrastructure related services are included in the infrastructure costs in different organisations.
2. Based on the survey results and cost model, approximate and evaluate the overall cost of the entire European HTC and HPC infrastructures.
3. Compare these results with commercial leased and on-demand offerings, such as Amazon services (EC2, S3 and “HPC on the cloud”), both in terms of cost and the differences in the various aspects these services when compared with the dedicated HTC and HPC e-Infrastructures.
4. Communicate the results to the wider European e-Infrastructure community through active dissemination, contributions to policy formation and organisation of dedicated workshops.