面向配备分布式储能的数据中心的概率功率供应理论
A Theory of Probabilistic Power Provisioning for Data Centers with Distributed Energy Storage
- The Ohio State University（俄亥俄州立大学）
- Massachussetts Institute of Technology（麻省理工学院）
机构由 AI 辅助整理，请以论文原文为准。
AI总结：
本文针对配备分布式储能的数据中心的电力供应问题，提出概率框架，划分两种运行模式并引入等效功率概念，经三类生产数据中心工作负载验证，为下一代AI数据中心提供规模设计原则。
AI中文摘要：
AI工作负载不断增长的电力需求及其波动性，使得电力输送成为数据中心运行的关键约束。分布式储能可降低支撑随机负载所需的电力容量，但其效益根本上取决于需求的统计特性和时间尺度。本文开发了一种概率框架，共同刻画供应电力、储能容量和透支概率。研究表明，储能辅助的电力供应可分为两种运行模式：在小电池区域，透支由短期需求偏移主导，储能可近乎线性地降低所需电力裕度；在大电池区域，透支源于较长时间跨度内的持续需求波动，所需裕度随储能增加呈现边际收益递减。针对该区域，本文引入了等效功率（effective power），这是等效带宽（effective bandwidth）的类似概念，可捕捉需求的时间统计特性，对所需电力给出渐近紧的刻画。本文进一步量化了时间相关性和空间聚合如何影响储能需求和统计复用增益，并将分析扩展到具有多个需求时间尺度的负载。最后，本文使用来自三个生产数据中心的电力需求轨迹评估该框架，这些数据中心覆盖HPC、GPU训练和云服务工作负载。尽管这些工作负载具有异构、循环平稳和多模态特性，但实测工作负载表现出预测的运行模式，且一个简单的四参数两状态模型可捕捉其储能-电力权衡的动态。所得框架为下一代AI数据中心的储能辅助电力供应提供了概率基础和实用的规模设计原则。
英文摘要：
The growing power demands and variability of AI workloads make electrical power delivery a critical constraint in data-center operation. Distributed energy storage can reduce the power capacity required to support stochastic loads, but its benefits depend fundamentally on the statistics and time scales of demand. This paper develops a probabilistic framework that jointly characterizes provisioned power, energy-storage capacity, and the probability of overdraw. We show that storage-assisted provisioning separates into two operating regimes. In the Small Battery Region, overdraw is dominated by short-lived demand excursions and storage provides nearly linear reductions in the required power margin. In the Large Battery Region, overdraw results from sustained demand fluctuations over longer spans of time, and the required margin exhibits diminishing returns with storage. For this regime we introduce effective power, an analogue of effective bandwidth that captures the temporal statistics of the demand and gives an asymptotically tight characterization of the required power. We further quantify how temporal correlation and spatial aggregation affect storage requirements and statistical multiplexing gains, and extend the analysis to loads with multiple demand time scales. Finally, we evaluate the framework using power-demand traces from three production data centers spanning HPC, GPU-training, and cloud-service workloads. Despite their heterogeneous, cyclo-stationary and multi-modal behavior, the measured workloads exhibit the predicted regimes, and a simple four-parameter two-state model captures the dynamics governing their storage-power tradeoffs. The resulting framework provides both a probabilistic foundation and practical dimensioning principles for storage-assisted power provisioning in next-generation AI data centers.
