摘要
随着云计算和大数据技术的迅猛发展,数据中心的任务负载迅速增长。如何在确保任务服务质量的前提下,采取有效措施降低数据中心的能耗成本,特别是在电价波动的背景下进行优化,已成为一项重要挑战。针对数据中心在电价波动背景下的能耗和任务延迟问题,提出了一种面向电价型需求响应的多目标联合优化策略。该策略引入电价信号响应机制,通过动态调节来实现对数据中心服务器资源的管理及负载的合理分配,并采用M/M/n排队模型对任务延迟进行建模,运用多目标进化算法能够同时求得帕累托最优解。算例结果表明,提出的优化策略能够在不同电价时段有效降低数据中心的能耗成本和任务延迟时间。系统在高电价时段通过削减服务器负载显著减少能耗,在低电价时段充分利用资源优化运营成本,并在任务延迟方面实现显著控制,满足服务质量协议的要求。该策略为数据中心在不同负载条件下提供了高效、灵活的调度方案,实现了能耗与任务处理的平衡优化。
With the rapid development of cloud computing and big data technologies,the task load of data centers has been growing rapidly.How to take effective measures to reduce the energy consumption costs of data centers while ensuring the quality of task services,especially under the fluctuations in electricity prices,has become a significant challenge.To address the issues of energy consumption and task latency in data centers in the context of electricity price fluctuations,a price-driven demand response multi-objective joint optimization strategy is proposed.This strategy incorporates a price signal response mechanism and achieves dynamic adjustment of server resource management and load distribution in data centers.It uses the M/M/n queuing model to model task latency and employs a multi-objective evolutionary algorithm to simultaneously obtain Pareto optimal solutions.Experimental results show that the proposed optimization strategy effectively reduces both the energy consumption costs and task latency of data centers during different electricity price periods.The system significantly reduces energy consumption by cutting server load during high electricity price periods,fully utilizes resources to optimize operating costs during low electricity price periods,and achieves significant control over task latency,meeting the requirements of the service level agreement(SLA).This strategy provides an efficient and flexible scheduling solution for data centers under varying load conditions,achieving balanced optimization between energy consumption and task processing.
作者
张益飞
高赐威
严兴煜
ZHANG Yifei;GAO Ciwei;YAN Xingyu(School of Electrical Engineering,Shanghai University of Electric Power,Shanghai 200000,China;School of Electrical Engineering,Southeast University,Nanjing 210000,China)
出处
《供用电》
北大核心
2024年第12期47-53,共7页
Distribution & Utilization
基金
国家自然科学基金青年科学基金项目(52107077)。
关键词
需求响应
数据中心
能耗成本
延迟时间
负荷管理
demand response
data center
energy consumption cost
latency
load management
