Scholarlink

As the demand for economically efficient cloud computing services grows, optimizing VM usage costs and reducing computational expenses are essential for providing better user value. This paper introduces the proposed model, EcoSLA-RL, an algorithm based on reinforcement learning for virtual machine (VM) consolidation in data centers, validated through simulations. Our goal is to provide a robust, user cost-efficient solution by leveraging reinforcement learning for dynamic consolidation, focusing on minimizing computational costs while effectively addressing challenges related to fluctuating workloads and SLA compliance. EcoSLA-RL achieved computation cost reductions of over 10% compared to other algorithms, while maintaining an SLA violation rate as low as 0.053% in large-scale environments.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

EcoSLA-RL: A Reinforcement Learning-Based Approach for Optimizing SLA and Execution Costs in Cloud Data Centers

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Chang Wu,
  • Pei-Wei Tsai,
  • Muneeb Ul Hassan,
  • Na Guo,
  • Yang Lin

摘要

As the demand for economically efficient cloud computing services grows, optimizing VM usage costs and reducing computational expenses are essential for providing better user value. This paper introduces the proposed model, EcoSLA-RL, an algorithm based on reinforcement learning for virtual machine (VM) consolidation in data centers, validated through simulations. Our goal is to provide a robust, user cost-efficient solution by leveraging reinforcement learning for dynamic consolidation, focusing on minimizing computational costs while effectively addressing challenges related to fluctuating workloads and SLA compliance. EcoSLA-RL achieved computation cost reductions of over 10% compared to other algorithms, while maintaining an SLA violation rate as low as 0.053% in large-scale environments.