Abstract: Since the current cloud task scheduling algorithm in the cryptography cloud environment cannot achieve the target that tasks are processed in real-time, a real-time threshold scheduling method based on rolling optimization window was proposed. Firstly, a cryptography cloud service architecture was given by integrating the link of key calling into the process of cryptographic task; secondly, to realize real-time scheduling, a cryptographic task scheduler model based on the rolling window and a throughput analysis model which was used to obtain the real-time throughput data were established; finally, to meet the objective needs of high-speed cryptographic service for cloud tenants, a throughput threshold scheduling algorithm was proposed, which migrates virtual cipher machine in real-time according to the changes of real-time throughput relative to throughput threshold. The simulation results show that compared with the method without the rolling optimization window or virtual machine migration technology, the proposed method has characteristics of shorter task completion time and lower CPU utility, meanwhile the real-time throughput of it can be continuously kept in 70%-85% of the network bandwidth, thus verifying its effectiveness and real-time performance in the cryptography cloud environment.

Key words: cryptography cloud, task scheduling, rolling optimization, throughput, threshold

摘要： 针对当前云任务调度算法在密码云环境中无法实现任务实时处理的问题，提出一种基于滚动优化窗口的实时阈值调度方法。首先，将密钥调用环节融入密码任务流程中，提出一种密码云服务架构；其次，为实现任务的实时调度，构建基于滚动窗口的密码任务调度器模型和吞吐量分析模型，用于获得实时的吞吐量数据；最后，为满足云租户对高速密码服务的客观需求，提出吞吐量阈值调度算法，从而根据实时吞吐量相对于吞吐量阈值的变化情况实时迁移虚拟密码机。仿真结果表明，该方法与未采用滚动优化窗口或虚拟机迁移技术的方法相比，具有任务完成时间短、CPU占用率低的特点，且实时吞吐量能够持续保持在网络带宽的70%~85%，从而验证了其在密码云环境中的有效性和实时性。

关键词: 密码云, 任务调度, 滚动优化, 吞吐量, 阈值