The NESUS Action will focus on a cross-community approach of exploring system software and applications for enabling a sustainable development of future high-scale computing platforms. In details, the Action will work in the following scientific tasks:
- First, the current state-of-the-art on sustainability in large-scale systems will be studied. The Action will strive for continuous learning by looking for synergies among HPC, distributed systems, and big data communities in cross cutting aspects like programmability, scalability, resilience, energy efficiency, and data management.
- Second, the Action will explore new programming paradigms, runtimes, and middlewares to increase the productivity, scalability, and reliability of parallel and distributed programming.
- Third, as failures will be more frequent in ultrascale systems, the Action will explore approaches of continuous running in the presence of failures. The Action plans to find synergies between resilient schedulers that handle errors reactively or proactively, monitoring and assessment of failures, and malleable applications that can adapt their resource usage at runtime.
- Fourth, future scalable systems will require sustainable data management for addressing the predicted exponential growth of digital information. The Action plans to explore synergistic approaches from traditionally separated communities to reform the handling of the whole data life cycle, in particular: restructure the Input/Output (I/O) stack, advance predictive and adaptive data management, and improve data locality.
- Fifth, as energy is a major limitation for the design of ultrascale infrastructures, the Action will address energy efficiency of ultrascale systems by investigating, promoting, and potentially standardizing novel metrics for energy monitoring and profiling, modelling, and simulation of energy consumption and CO2 emission, eco-design of ultrascale components and applications, energy-aware resource management, and hardware/software codesign.
- Finally, the Action will identify applications, high-level algorithms, and services amenable to ultrascale systems and investigate the redesign and reprogramming efforts needed for applications to efficiently exploit ultrascale platforms, while providing sustainability.