首页> 外文会议>International Conference on Computational Collective Intelligence >Applying Supervised Machine Learning to Predict Virtual Machine Runtime for a Non-hyperscale Cloud Provider
【24h】

Applying Supervised Machine Learning to Predict Virtual Machine Runtime for a Non-hyperscale Cloud Provider

机译:应用监督机器学习来预测非超大规模云提供商的虚拟机运行时间

获取原文

摘要

Cloud computing offers an online, on-demand and pay-as-you-go access to computing resources. The cloud enables users to adjust their consumption to their needs. Users deploy their application code, libraries and operating systems on the provider's hardware. The resources can be allocated under the form of virtual machines (VMs). Predicting the runtime of VMs can be useful to optimize the resource allocation. We propose a formulation of this objective as a multi-class classification problem by using as much features as available when launching a VM. Experimentation carried out on real traces from the public cloud provider Outscale show that the inclusion of features extracted from tags, which are freely-typed pieces of text used to describe VMs for human operators, improve the model performance.
机译:云计算提供对计算资源的在线,按需和按需购买的访问。云使用户可以根据自己的需求调整消费量。用户在提供商的硬件上部署他们的应用程序代码,库和操作系统。可以以虚拟机(VM)的形式分配资源。预测VM的运行时间对于优化资源分配很有用。我们建议通过启动虚拟机时使用尽可能多的功能,将该目标表述为多类分类问题。对来自公共云提供商Outscale的真实痕迹进行的实验表明,包含从标签中提取的功能(这些内容是用于为操作员描述VM的自由键入的文本片段),可以改善模型性能。

著录项

相似文献

  • 外文文献
  • 中文文献
  • 专利
获取原文

客服邮箱:kefu@zhangqiaokeyan.com

京公网安备:11010802029741号 ICP备案号:京ICP备15016152号-6 六维联合信息科技 (北京) 有限公司©版权所有
  • 客服微信

  • 服务号