A Vertical Approach to Designing and Managing Sustainable Heterogeneous Edge Data Centers

TL;DR

This paper introduces a comprehensive vertical integration approach for designing and managing sustainable Edge Data Centers, focusing on reducing both operational and embodied carbon emissions through innovative design, system, and runtime strategies.

Contribution

It presents a novel holistic methodology that integrates architecture, system, and runtime layers for carbon-aware optimization of EDCs, challenging existing paradigms.

Findings

Carbon-aware accelerator designs reduce embodied carbon.

Adaptive resource utilization minimizes operational emissions.

Dynamic scheduling aligns energy use with carbon intensity.

Abstract

The increasing demand for Artificial Intelligence (AI) computing poses significant environmental challenges, with both operational and embodied carbon emissions becoming major contributors. This paper presents a carbon-aware holistic methodology for designing and managing sustainable Edge Data Centers (EDCs), based on three design principles that challenge the state-of-the-art optimization paradigms. Our approach employs vertical integration across the architecture, system, and runtime layers, balances operational and embodied carbon emissions while considering EDC performance as a co-optimization objective, rather than a constraint. At the architecture level, we propose carbon-aware and approximate accelerator designs to reduce embodied carbon. At the system level, we enhance resource utilization and adapt to real-time carbon intensity variations to minimize operational emissions. Finally, at the runtime level, we develop dynamic scheduling frameworks that adjust execution, based on energy constraints and carbon intensity.