信息运维综合监管系统中的内存数据库集群技术  

Cluster Technique of In-Memory Database in Integrated Information Management System

作　　者：梁鸿健[1] 胡游君[2] 唐海荣[2] 

机构地区：[1]广东电网公司深圳供电局,广东深圳218001 [2]国网电力科学研究院,江苏南京210003

出　　处：《中国电子商情（通信市场）》2011年第6期181-186,共6页

摘　　要：信息系统一般都采用传统的关系型数据库，传统的关系型数据库管理系统主要强调维护数据的完整性、一致性、稳定性，一般采用物理磁盘存储的方式。由于磁盘存取、内外存的数据传输、缓冲区管理、排队等待及锁的延迟等缺点使得数据库I／O不能满足信息运维综合监管系统中各类实时数据处理、响应的应用需求，如果将整个数据库管理系统放入内存，使每个事务I／O消耗变得极短，则为高速数据处理和存储的应用场景提供了有力的支持。为了保证数据的安全性，传统的关系型数据库管理系统基本上都采用多节点集群方式来保证数据的多节点存储和数据安全。如果内存数据库管理系统也能实现多节点集群部署，则能有效的解决内存数据库系统存在数据库系统稳定性、数据安全性等问题，从而符合信息运维综合监管系统对内存数据库高速数据处理的应用需求。本文提出了一种支持互备、扩展、混合三种模式的内存数据库的集群技术，在提高内存数据库的稳定性和数据安全性的前提下发挥内存数据库的高速I／O优势。Generally speaking, IT system adopts traditional relational databases which emphasize on the completely, consistence and security of data stored in physical disc. Due to delay of disc I/O, data transmission, Operating system buffer manage, queue wait and system lock etc., database I/O fail to satisfy the application of real time data process and response in integrated information management system. Putting the whole database management system in memory makes every transaction I/O consumption extremely short, which also provides strong support to the application of high speed data processing and storing. To ensure the security of the data, traditional relational databases usually employ multi-node cluster to make sure multi-node storage and security of the data. Accordingly, multi-node cluster in-memory data management system would effectively ensure the stability and security of database system, thus satisfies the need of integrated information management system to high speed data handle in in-memory database. The paper put forward to a cluster technique of in-memory database contains backup, extend, mix mode, to take advantage of high speed data I/O of in-memory database while enhancing the stability of in-memory database and data security.

关 键 词：内存数据库 数据库集群 信息运维 实时监控 

分 类 号：TP311.13[自动化与计算机技术—计算机软件与理论]