Scholarlink

<p>This study investigates joint task scheduling and virtual machine placement (JTSVMP) in heterogeneous cloud data centres, where task-to-VM assignment and VM-to-physical-host placement must be optimized together to reduce execution delay, cost, energy consumption, resource fragmentation, and SLA violation. Although the Java Macaque Algorithm (JMA) was originally proposed as a general meta-heuristic, this work adapts it to JTSVMP through a joint task–VM–PH encoding, deadline- and capacity-aware initialization, Pareto-rank and grid-assisted candidate ordering, feasibility restoration, and consolidation-aware physical-host activation. The proposed method was evaluated in CloudSim using NASA, HPC2N, and Synthetic workloads. Each dataset–algorithm–task-size configuration was repeated 30 times and compared with EMO-TS, HGHHC, GA, HACOS, and IGSO. The results show that JMA reduces makespan, execution time, execution cost, energy consumption, active physical hosts, resource wastage, degree of imbalance, runtime overhead, and SLA violation while improving resource utilization. Relative to the closest energy-aware baseline, EMO-TS, JMA reduces makespan, execution time, and execution cost by approximately 2.15% and energy consumption by approximately 0.79%. These results indicate that the proposed JMA-based JTSVMP framework provides balanced scheduling and placement improvements without increasing service-feasibility risk.</p>

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

Novel meta-heuristic optimization approach for joint task scheduling and virtual machine placement using Java Macaque Algorithm

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年6月23日

查看全文
  • Dinesh Karunanidy,
  • Nandhini Mahadevan,
  • Mamoon Rashid,
  • Sultan S. Alshamrani,
  • Jasem Almotiri

摘要

This study investigates joint task scheduling and virtual machine placement (JTSVMP) in heterogeneous cloud data centres, where task-to-VM assignment and VM-to-physical-host placement must be optimized together to reduce execution delay, cost, energy consumption, resource fragmentation, and SLA violation. Although the Java Macaque Algorithm (JMA) was originally proposed as a general meta-heuristic, this work adapts it to JTSVMP through a joint task–VM–PH encoding, deadline- and capacity-aware initialization, Pareto-rank and grid-assisted candidate ordering, feasibility restoration, and consolidation-aware physical-host activation. The proposed method was evaluated in CloudSim using NASA, HPC2N, and Synthetic workloads. Each dataset–algorithm–task-size configuration was repeated 30 times and compared with EMO-TS, HGHHC, GA, HACOS, and IGSO. The results show that JMA reduces makespan, execution time, execution cost, energy consumption, active physical hosts, resource wastage, degree of imbalance, runtime overhead, and SLA violation while improving resource utilization. Relative to the closest energy-aware baseline, EMO-TS, JMA reduces makespan, execution time, and execution cost by approximately 2.15% and energy consumption by approximately 0.79%. These results indicate that the proposed JMA-based JTSVMP framework provides balanced scheduling and placement improvements without increasing service-feasibility risk.