Energy-Aware Edge Orchestration: Turning Edge Sites into Grid-Supporting Assets

With rapid advances in artificial intelligence (AI) algorithms hosted in cloud data centres, there is increasing demand to also support AI algorithms closer to where data is generated and used. Such edge computing infrastructure does not just support computation and data storage; it can also help the electricity grid when tasks are scheduled using information about local energy availability, electricity prices, and the carbon intensity of the power being used, that is, how much CO2 is emitted per kilowatt-hour of electricity supplied on the grid. This chapter reframes energy-aware edge orchestration as a set of practical scheduling patterns and the data and control interfaces they require. We motivate the need using familiar grid dynamics such as steep ramping, then present a reference architecture and the signals and standards that enable orchestration decisions in real-world applications. We position this narrative against our prior work on edge energy orchestration, and we close with practical considerations for experts integrating carbon, price, and device telemetry into schedulers.