论文部分内容阅读
日志数据管理系统是最重要的云服务基础设施之一。重要日志数据缺失将造成相应日志分析与决策的片面性和不准确性。然而日志数据采集能力越强,日志采集的运行期开销就越大,海量日志数据的管理与分析就越耗时,对整个云服务环境的系统性能造成不可忽视的影响。针对如何采集必要的日志数据同时尽可能降低其运行期开销的问题,文章首先提出日志采集粒度的概念,然后设计并编程实现一个面向云计算的粒度自配置日志采集平台。其中,平台构成模块包括:日志采集工具、存储日志采集粒度规则和事实的知识库;基于规则动态增加或关闭相关日志数据采集模块的推理机;相应的图形界面,包括用于添加或修改知识库规则的管理界面和直观查看日志数据的用户界面。最后,初步的案例学习结果表明了平台的有效性。
Log data management system is one of the most important cloud service infrastructure. The lack of important log data will result in the one-sidedness and inaccuracy of the corresponding log analysis and decision. However, the stronger the log data acquisition capability, the greater the log acquisition runtime costs, the more time-consuming the management and analysis of massive log data, and the non-negligible impact on the system performance of the entire cloud service environment. In order to collect the necessary log data and reduce the overhead of its operation as much as possible, this paper first proposes the concept of log collection granularity and then designs and programs a granular log collection platform for cloud computing. The platform component module includes: a log collection tool, a knowledge base that stores log collection granularity rules and facts, an inference engine that dynamically adds or closes related log data collection modules based on rules, and a corresponding graphical interface including a database for adding or modifying a knowledge base A rule management interface, and a user interface to visually view log data. Finally, the initial case study shows the effectiveness of the platform.