The PULSE challenge, shared by Inria and Qarnot computing, aims to develop and promote best practices in geo-distributed hardware and software infrastructures for intensive computing with a reduced environmental footprint.

The aim of this postdoc is to better quantify the difference in environmental footprint of different models of geo-distributed computing, so as to be able to better manage the induced impacts (according to a multi-criteria approach). Ultimately, we hope to be able to recommend best practice hardware and software architectures to drastically reduce impacts while offering the best quality of service to end-users.

This project aims to develop two models to compare the environmental impact of a centralised and a decentralised computing service.

The two models should be based on the same assumptions and follow methodologies that are as close as possible in order to achieve directly comparable results.

The project will not be limited to modelling the operational carbon footprint of the service, but will also include :

• all peripheral services requested by the calculation, i.e. storage and network services, etc.
• all the peripheral infrastructures concerned by the calculation, i.e. the land of the data centre, data transport, etc.
• the analysis of the life cycle of the system,
• the entire environmental footprint: in addition to carbon and greenhouse gas emissions, the impact on rare metals, on water, on soil pollution, etc.
• more qualitative data to develop a case for the most virtuous model

The researcher will have to understand these different subjects through a consequent bibliographical research. The researcher will then have to work on the modelling of each of these paradigms using the most suitable tools. It will be potentially desirable to propose several qualities of models, for example an advanced model and a simpler model. In all cases, it will be essential to produce a model that is easy to use and modify, a priori in the form of an Excel file.

In addition, it will be necessary to develop a more qualitative argument for the most virtuous model.