By live migration technology, multiple virtual machines (VMs) can be consolidated into a fewer physical servers and the idle ones can be shut down or switched to low-power mode, thus reducing the energy consumption of cloud data centers. However, live migration can result in performance degradation of migrated VMs, or even interrupting their services. At the same time, live migration can also aggravate the overheads of data transmissions and produce additional energy consumption in cloud data centers. All these negative influences belong to migration cost (MC) caused by VM migration, which becomes an important cost factor that can’t be ignored. Otherwise, another important concern, remaining runtime of the migrated VM, also has influence on the efficiency of VM consolidation, which is not well addressed as well. This paper investigates MC-aware VM consolidation problem and formulates the problem as a multi-constraint optimization model by considering migration cost and remaining runtime of VMs. Based on the proposed model, a heuristic algorithm, called MC-aware VM consolidation (MVC) algorithm, is developed. Finally, based on a real-world cloud trace, we conduct extensive experimental studies to verify the validity of the proposed model and algorithm. Experimental results show that, compared with some popular algorithms, MVC algorithm can effectively decrease the migration cost and, at the same time guarantee the energy consumption within a certain low level.

中文翻译：

---

考虑剩余运行时间的云环境下迁移成本和能量感知虚拟机整合

通过实时迁移技术，可以将多个虚拟机（VM）整合到更少的物理服务器中，将空闲的虚拟机关闭或切换到低功耗模式，从而降低云数据中心的能耗。但是，实时迁移可能会导致迁移的虚拟机性能下降，甚至中断其服务。同时，实时迁移也会加重云数据中心的数据传输开销，产生额外的能耗。所有这些负面影响都属于VM迁移带来的迁移成本（MC），成为不可忽视的重要成本因素。否则，另一个重要问题，即迁移后的 VM 的剩余运行时间，也会影响 VM 整合的效率，这一点也没有得到很好的解决。本文研究了 MC-aware VM 整合问题，并通过考虑迁移成本和 VM 的剩余运行时间将该问题表述为多约束优化模型。基于所提出的模型，开发了一种启发式算法，称为 MC 感知 VM 合并 (MVC) 算法。最后，基于真实世界的云迹，我们进行了广泛的实验研究，以验证所提出的模型和算法的有效性。实验结果表明，与一些流行的算法相比，MVC算法可以有效降低迁移成本，同时保证能耗在一定的低水平。开发了一种启发式算法，称为 MC 感知 VM 合并 (MVC) 算法。最后，基于真实世界的云迹，我们进行了广泛的实验研究，以验证所提出的模型和算法的有效性。实验结果表明，与一些流行的算法相比，MVC算法可以有效降低迁移成本，同时保证能耗在一定的低水平。开发了一种启发式算法，称为 MC 感知 VM 合并 (MVC) 算法。最后，基于真实世界的云迹，我们进行了广泛的实验研究，以验证所提出的模型和算法的有效性。实验结果表明，与一些流行的算法相比，MVC算法可以有效降低迁移成本，同时保证能耗在一定的低水平。