Sustainable AIGC Workload Scheduling of Geo-Distributed Data Centers: A Multi-Agent Reinforcement Learning Approach

被引:2
作者
Zhang, Siyue [1 ,2 ]
Xu, Minrui [2 ]
Lim, Wei Yang Bryan [1 ,2 ]
Niyato, Dusit [2 ]
机构
[1] Alibaba NTU Singapore Joint Res Inst, Singapore, Singapore
[2] Nanyang Technol Univ, Sch Comp Sci & Engn, Singapore, Singapore
来源
IEEE CONFERENCE ON GLOBAL COMMUNICATIONS, GLOBECOM | 2023年
关键词
AI-generated content; Job scheduling; Green cloud computing; Multi-agent reinforcement learning;
D O I
10.1109/GLOBECOM54140.2023.10437617
中图分类号
TM [电工技术]; TN [电子技术、通信技术];
学科分类号
0808 ; 0809 ;
摘要
Recent breakthroughs in generative artificial intelligence have triggered a surge in demand for machine learning training, which poses significant cost burdens and environmental challenges due to its substantial energy consumption. Scheduling training jobs among geographically distributed cloud data centers unveils the opportunity to optimize the usage of computing capacity powered by inexpensive and low-carbon energy and address the issue of workload imbalance. To tackle the challenge of multi-objective scheduling, i.e., maximizing GPU utilization while reducing operational costs, we propose an algorithm based on multi-agent reinforcement learning and actor-critic methods to learn the optimal collaborative scheduling strategy through interacting with a cloud system built with real-life workload patterns, energy prices, and carbon intensities. Compared with other algorithms, our proposed method improves the system utility by up to 28.6% attributable to higher GPU utilization, lower energy cost, and less carbon emission.
引用
收藏
页码:3500 / 3505
页数:6
相关论文
共 17 条
  • [1] Adami Davide, 2013, 2013 IEEE International Conference on Communications (ICC), P2578, DOI 10.1109/ICC.2013.6654923
  • [2] [Anonymous], FIN TUN STABL DIFF W
  • [3] Electricity Intensity of Internet Data Transmission: Untangling the Estimates
    Aslan, Joshua
    Mayers, Kieren
    Koomey, Jonathan G.
    France, Chris
    [J]. JOURNAL OF INDUSTRIAL ECOLOGY, 2018, 22 (04) : 785 - 798
  • [4] Gao W., 2022, ARXIV220511913
  • [5] Haarnoja T., 2018, ARXIV
  • [6] huggingface, Masked Language Modeling
  • [7] Optimal data placement strategy considering capacity limitation and load balancing in geographically distributed cloud
    Li, Chunlin
    Cai, Qianqian
    Youlong, Lou
    [J]. FUTURE GENERATION COMPUTER SYSTEMS-THE INTERNATIONAL JOURNAL OF ESCIENCE, 2022, 127 : 142 - 159
  • [8] Luccioni A.S., 2022, Estimating the Carbon Footprint of BLOOM, a 176B Parameter Language Model
  • [9] Applications of Deep Reinforcement Learning in Communications and Networking: A Survey
    Luong, Nguyen Cong
    Hoang, Dinh Thai
    Gong, Shimin
    Niyato, Dusit
    Wang, Ping
    Liang, Ying-Chang
    Kim, Dong In
    [J]. IEEE COMMUNICATIONS SURVEYS AND TUTORIALS, 2019, 21 (04): : 3133 - 3174
  • [10] A review on job scheduling technique in cloud computing and priority rule based intelligent framework
    Murad, Saydul Akbar
    Muzahid, Abu Jafar Md
    Azmi, Zafril Rizal M.
    Hoque, Md Imdadul
    Kowsher, Md
    [J]. JOURNAL OF KING SAUD UNIVERSITY-COMPUTER AND INFORMATION SCIENCES, 2022, 34 (06) : 2309 - 2331