摘要
提出一种分布式实时数据系统中的基于动态索引策略的存取定位机制。该机制结合了改进的一致性哈希算法与"虚拟Token空间"机制。在该机制基础上,针对流程工业领域测点数据的特征和需求,用大量的廉价服务器实现了实时性高、吞吐量大、可扩展性好的分布式实时数据库系统。该数据库系统具有测点数据写入位置的自寻找、服务器集群内部状态变更时测点数据位置的自切换、数据查询的自优化、服务器负载自均衡等优势。该系统封装了分布式系统的各种瓶颈问题,使用户不需要关心系统内部的机制,可以像使用单体数据库系统一样直接向系统读写数据,而该系统性能又是单体数据库的叠加。通过在原型系统中测试验证,结果表明该机制能够达到设计预期目标和性能要求。
A storage and inquiry location mechanism in distributed real-time data system based on dynamic index strategy is proposed.Combining the improved consistent hash algorithm with a ＂Virtual Token Space＂ mechanism and needs of the measuring point data in the field of process industries,the mechanism achieves a highly real-time,large throughput per second and good scalability distributed real-time database systems.The database system has advantages of point-data self locating,location self switching when the internal state of the server cluster changed,self-optimizing of data inquiry,self balancing of servers＂ load.The system encapsulates a variety of bottlenecks problems in distributed systems,so that users do not need to care about the internal mechanisms of systems.Users could read and write data directly toward the system like using a single-server database,but the system performance is the superposition of the single database.
出处
《工业控制计算机》
2013年第11期4-7,9,共5页
Industrial Control Computer
