中国科学院软件研究所机构知识库

综合信息系统中分系统越来越多，而且日益复杂。各个分系统之间数据交互的要求也不断增加，同时各种历史数据爆炸性增加。实现综合信息系统中的数据实时、高效地分发，以及各个分系统的海量历史数据存储变得日益重要。

本文针对综合信息系统集成中通信的实时、高效性问题和海量历史数据存储问题，在对数据分发服务进行深入分析和研究的基础上，设计了以数据为中心的数据分发服务模型，并结合关系型数据存储和非关系型数据存储，提出一种海量历史数据的存储、管理方法。

首先针对现有中间件通信模型不能满足综合信息系统越来越复杂的应用需求，比如通信的实时性、快捷性和灵活性的问题。基于对象管理组织发布的数据分发服务规范，设计以数据为中心的适合分布式实时通信的数据分发服务模型，保证数据高效、可靠地分发。为了实时监控通信过程中实体的状态及其变化，在原有的OpenDDS分发服务中加入监控接口设计，实时监测发布、订阅的实体及其状态。

然后针对海量历史数据的存储和管理问题，采用分布式文件系统实现数据分布式存储管理，完成对海量数据的存储和管理操作。利用关系型数据库SQL Server和非关系型数据库MongoDB设计了混合分布式存储架构HybridDB。

最后本文以船舶信息系统中的电力分系统、推进分系统、损管分系统和综合船桥分系统为例，搭建了一套基于数据分发服务的监控与存储仿真系统。为船舶信息系统集成中的通信实时、高效性需求和海量历史数据的有效存储、高效查询需求提供了很好的解决方案。

With the integrated information systems become much more complex, the requirement of data exchange between the various subsystems increase explosively. At the same time, a variety of historical data increase explosively.The real-time, efficient data distribution in the integrated information system is becoming increasingly important, as well as the massive historical data storage of various subsystems.

Against the real-time, efficient data distribution and massive historical data storage twokey problems in the integrated information system, after analyzing and researching the specifications of Data Distribution Services, a data-centric data distribution service model is proposed. At the same time, the relational data storage and non-relational data storage are combined to propose a solution to store and manage the massive historical data.

Firstly, the existing middleware communication model cannot meet the increasingly complex application requirements of integrated information systems, such as real-time communication, speed and flexibility. For the assurance of distributing data in real-time and efficiently, a data-centric data distribution service model for distributed real-time communicationsystems is designed. It isbased on the specification of Data Distribution Services. In order to monitor the status and changes of the entity in the communication process, a monitoring interface model is proposed and insertedinto the original OpenDDS distribution service.

Secondly, in order to implement the storage and management of massive historical data, a distributed storage management based on distributed file system is proposed. It combines SQL Server and MongoDB to implement a hybrid distributed storage architecture which names HybridDB.

To verify the solution, we design separate simulation experiments based on ship information system, which includes electricity sub-system, propelling sub-system, damage control sub-system and integrated bridge subsystem.The experiments prove that it provides a perfect solution for the real-time communication and the effective storage of historical data during doing the integration ofvarious subsystems for the ship information system.