【摘要】 容器化为云计算和边缘计算应用提供了一种更加便捷和快速的资源供应与管理方法。边缘环境下资源有限，现有的容器调度算法大多只考虑单个应用，没有考虑多应用之间的资源竞争问题。因此提出一种针对边缘环境的多应用容器调度方法，该方法利用长短期记忆网络对地域分布负载进行预测，采用深度强化学习与环境不断交互获取应用的最佳容器数量，并根据启发式规则解决应用间的资源冲突问题。实验结果表明，与已有方法相比，该方法具有更低的请求违反率和资源消耗成本。更多还原

【Abstract】 Containerization provides a more convenient and faster resource provisioning and management method for cloud and edge computing applications. Resources are limited in the marginal environment,and most of the existing container scheduling algorithms only consider a single application,without considering the resource competition among multiple applications. Therefore,a multi-application container scheduling method for marginal environment is proposed. This method uses long and short term memory network to predict the regional distribution load,uses deep reinforcement learning and environment to continuously interact to obtain the optimal number of containers for applications,and solves the resource conflict between applications according to heuristic rules. Experimental results show that the proposed method has lower request violation rate and resource consumption cost than the existing methods.更多还原