服务创新平台弹性负载均衡机制的研究与实现

发布时间：2018-11-10 08:08

【摘要】：在服务创新平台上部署的部分应用在实际运行过程中面临着巨大的业务压力,这些应用本身也有着对高并发业务量的测试和验证需求,因此要求平台实现负载均衡服务为其提供运行保障和实验环境。传统的静态负载均衡机制需要应用事先指定自身所需要的负载机数目,不适合业务量波动比较大的应用。服务创新平台需要的负载均衡服务应该是一种更为灵活的形态,应用能实时根据自身的业务压力大小来调整负载资源,该种形态就是弹性负载均衡。弹性负载均衡机制根据监控信息来弹性伸缩机群,这种方式是被动应激式的,有一定的滞后性,服务创新平台需要尽可能的减短这段滞后时间。应用的运维人员在进行负载均衡配置时需要进行繁琐的配置工作,服务创新平台希望能自动化这个过程来提升运维人员效率。为了实现上述目标,本论文在服务创新平台上研究实现了一种自动化的智能弹性负载均衡机制,具体包括：实现弹性负载均衡服务,系统可根据监控反馈动态调整应用的负载资源,具体涉及到Docker技术的应用、底层网络的调配、OpenStack源代码的修改和Libvirt的监控等;实现分析预测机制和资源池模块来解决弹性伸缩过程中的滞后性问题,具体牵涉到决策树分类算法的应用、历史信息的存储等;实现前端异步交互模块为应用运维人员提供负载均衡管理界面,具体牵涉到异步处理机制的实现、浏览器的适配等。论文在最后部分对智能弹性负载均衡机制进行实验和测试,验证了该机制能在平台上有效运行。实验结果表明应用采用该机制后对业务压力的应对能力有较大提升,应用平均响应时间下降了34%左右。
[Abstract]:Some of the applications deployed on the service innovation platform are facing enormous business pressure in the actual operation process. These applications themselves also have the test and verification requirements for high concurrent traffic. Therefore, the platform is required to implement load balancing services to provide it with operational support and experimental environment. The traditional static load balancing mechanism needs to specify the number of load machines it needs in advance, so it is not suitable for applications with large traffic fluctuation. The load balancing service needed by the service innovation platform should be a more flexible form. The application can adjust the load resource according to its own business pressure in real time. This kind of form is flexible load balancing. The flexible load balancing mechanism is based on the monitoring information to stretch the cluster. This way is passive stress, there is a certain lag, service innovation platform needs to reduce the lag time as soon as possible. The application operator needs to carry on the tedious configuration work when carries on the load balancing configuration, the service innovation platform hopes to automate this process to enhance the operation and maintenance personnel efficiency. In order to achieve the above goal, this paper studies and implements an automatic intelligent elastic load balancing mechanism on the service innovation platform, including: realizing the flexible load balancing service, The system can dynamically adjust the load resources of the application according to the monitoring feedback, including the application of Docker technology, the deployment of the bottom network, the modification of OpenStack source code and the monitoring of Libvirt. The implementation of analysis and prediction mechanism and resource pool module to solve the problem of lag in the elastic expansion process, which involves the application of decision tree classification algorithm, the storage of historical information, etc. The implementation of the front-end asynchronous interaction module provides the load balancing management interface for the application operators, which involves the implementation of asynchronous processing mechanism, the adaptation of browsers, and so on. In the last part of the paper, the intelligent elastic load balancing mechanism is tested and tested, and it is verified that the mechanism can work effectively on the platform. The experimental results show that the ability to deal with the business pressure has been greatly improved and the average response time of the application has decreased by about 34%.
【学位授予单位】：北京邮电大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP393.09

【相似文献】