The virtual machine (VM) allocation problem is one of the main issues in cloud data centers. This article proposes a new metaheuristic method to optimize joint task scheduling and VM placement (JTSVMP) in cloud data center. The JTSVMP problem, though composed of two parts, namely task scheduling and VM placement, is treated as a joint problem to be resolved by using metaheuristic optimization algorithms (MOAs). The proposed co-optimization process aims to schedule task into the VM which has the least execution cost within deadline constraint and then to place the selected VM on most utilized physical host (PH) within capacity constraint. To evaluate the performance of our proposed co-optimization process, we compare the performances of two different scenarios, i.e., task scheduling algorithms and integrateion co-optimization of task scheduling and VM placement using MOAs, namely the basic glowworm swarm optimization (GSO), moth-flame glowworm swarm optimization (MFGSO) and genetic algorithm (GA). Simulation results show that optimizing joint task scheduling and VM placement leads to better overall results in terms of minimizing execution cost, makespan and degree of imbalance and maximizing PHs resource utilization.

中文翻译：

---

云数据中心联合任务调度和虚拟机放置的元启发方法

虚拟机（VM）分配问题是云数据中心的主要问题之一。本文提出了一种新的元启发式方法来优化云数据中心中的联合任务调度和虚拟机放置（JTSVMP）。 JTSVMP问题虽然由任务调度和VM放置两部分组成，但被视为一个联合问题，需要使用元启发式优化算法（MOA）来解决。所提出的协同优化过程旨在将任务调度到截止日期约束内执行成本最低的VM中，然后将所选VM放置在容量约束内最常用的物理主机（PH）上。为了评估我们提出的协同优化过程的性能，我们比较了两种不同场景的性能，即任务调度算法和使用 MOA 的任务调度和虚拟机放置的集成协同优化，即基本的萤火虫群​​优化（GSO），蛾焰萤火虫群优化（MFGSO）和遗传算法（GA）。仿真结果表明，优化联合任务调度和虚拟机布局在最小化执行成本、完工时间和不平衡程度以及最大化 PH 资源利用率方面带来了更好的总体结果。