基于OpenStack框架的数据中心资源优化问题研究

Data Center Resource Optimization Based on OpenStack Framework

多种多样的应用混合搭建在数据中心上,给网络应用环境带来了巨大变化,拥塞的网络性能影响着用户体验,甚至会造成客户流失,如何合理的调度任务并控制延迟时间阈值成为亟待解决的难题。在OpenStack虚拟框架基础上,提出一种结合马尔可夫链的数据中心资源优化算法,建立了一种任务接入阶段的优化控制机制,进而对数据中心资源进行优化分配,以保证交互负载流的延迟时间在客户容忍度阈值之内,从而提升客户的用户体验。通过实验表明,硬件配置越高响应时间越少,但其提升的性能优势不显著,所提算法能够有效缩短交互负载流请求情况下的延迟时间,而对非交互负载流的延迟时间影响不明显,且在处理高并发请求系统中具有较好的拓展性,为云服务器的资源优化提供了解决思路。

Hybrid construction of diverse applications in the data center brings great changes to the network application environment. The congested network performance affects the user experience and even causes customer loss. How to properly schedule tasks and control the delay time threshold becomes a challenge demanding urgent solution. Based on the OpenStack virtual framework, a data center resource optimization algorithm based on Markov chain is proposed. An optimization control mechanism is established in the task access phase, and then the data center resources are optimally allocated to ensure that delay time of interactive load flow is within the customer tolerance threshold, thereby improving customer's user experience. Experiments show that the response time is lower under higher hardware configuration, but the performance advantage of the improvement is not significant. The proposed algorithm can effectively shorten the delay time under the interactive load flow request, but exerts insignificant effect on delay time of non-interactive load flow. Moreover, it has better scalability in processing high concurrent request system, which provides a solution for resource optimization of cloud server.