Abstract: To fulfill the requirement of data warehouse construction for equipment status analysis,an efficient construction technology for distributed real-time data warehouse was proposed based on dimensional modeling theory,distributed message queue service and distributed streaming computation frame.Compared to the traditional data warehouse,an improved method of data life-cycle for distributed real-time data warehouse was proposed.The construction technology for distributed real-time data warehouse with features of multi-scenario,efficient iteration,high expansibility of performance and high availability of data was addressed.To support the management of data cloud platform in distributed real-time data warehouse,three current solutions of automatic operation and maintenance were evaluated,and the integration technology of data cloud platform was discussed.

Key words: equipment industry, status data, distributed cluster, real-time data warehouse, big data platform management

摘要： 面向大型装备状态分析的数据仓库建设需求,在数据仓库维度建模理论、分布式消息队列、分布式流式计算的基础上,提出一种快速构建分布式实时数据仓库的方法。相比于传统数据仓库,提出了分布式实时数据仓库在数据生命周期的改进方法。研究提出一种面向多数据场景的、可快速迭代的、具有高扩展性与数据可靠性的分布式实时数据仓库构建方法。为支撑分布式实时数据仓库的数据云平台管理,总结了现有的三种集群自动化运维方法,并提出了对多种数据云平台集成的方法。

关键词: 装备行业, 状态数据, 分布式集群, 实时数据仓库, 大数据平台管理