Energy Proportional Edge Computing Infrastructure – Francesc Guim, Intel Corp
Most companies are pledging net-zero carbon in next 2 decades, which will require ~ 40-50% reduction by 2030. In addition to "net-zero" goals, edge computing infrastructure being deployed face unique challenges to balance latency, security, distributed management, thermal ambients for outdoors equipment and increasing energy costs due to climate and socioeconomic events. To overcome these distributed edge infrastructure sustainability challenges, we will present "Energy Proportional Edge Computing Infrastructure" architecture, framework and a blueprint for sustainable "Net Zero" deployment using energy proportionality and carbon aware workload distribution and management using sustainability optimization techniques to balance workload requirements with energy consumption and carbon footprint. 1) Time shifting: moving functions to time-of-day based on energy availability with lowest carbon intensity; 2) Space shifting: moving functions to another location where energy with lower carbon intensity is available; 3) Energy Proportional optimizations: balancing function’s service level objective and managing system power states to match available energy and Carbon intensity proportionality.
