营销类多媒体信息发布系统终端自动化运维

摘要

With the increasing of terminal scale and deployment time, the information publish system for marketing multimedia is hard to maintain because the traditional single server model has the limits of storage and performance scalability, and the lack of real-time monitoring schema, fault alarm and handling mechanism. This paper uses the basic services such as cloud storage and computing provided by the industry of streaming media segmentation to redesign the architecture and realize linear expansion.It analyzes the business characteristics and maintain requirement of the information publish system, and determines the index system of monitoring terminals and the real-time monitoring scheme based on finite state machine. It reduces the time of error handling and improves the system operations. Then it focuses on the logic validity check of terminal player based on longest common sub-sequence and the evaluation model of video distortion based on frame rate and image quality, to realize the real-time monitoring forthe playing video quality. Experiments demonstrate that this solution is effective and acceptable.%随着终端规模的增加、系统部署时间的增长，营销类多媒体信息发布系统由于传统单服务器模型在存储、性能可扩展性的限制，实时监控及故障预警与处理机制的缺乏而难以运维。利用流媒体细分行业云提供的云存储、云计算等基础服务重新设计架构，实现线性可扩展；分析信息发布系统业务特色及运维需求，确定终端监控指标体系及基于有限状态机的实时监控方案，降低故障处理时间，提高系统可运维性；重点讲述基于字符串最长公共子序列的终端播放逻辑正确性检测，基于帧率、图像质量的视频失真评估模型，实现视频播放质量的实时监控。实验显示运维方案既能满足终端性能需求，又能达到运营预期。