基于NoSQL系统的组合索引技术研究

Study of Multiterm Indexing Techniques Based on NoSQL System

分布式NoSQL系统旨在提供大规模数据的高可用性,但缺乏内在的支持复杂查询的应用程序。传统的基于单一词汇倒排表的解决方案未达到良好的效果。因此,文中就文档型数据库在处理动态文档集时不支持多键作为主索引的缺点展开研究,提出了一种改进的组合索引方法。通过存储组合条件的倒列表,查询驱动机制可以从最近的查询记录中自适应地存储比较受欢迎的条件组合。该方法可以降低整体的带宽消耗,只需占用较少的存储资源等额外开销,明显改善了NoSQL系统的容量和响应时间。

The purpose of distributed NoSQL systems is to provide high availability for large- scale of data,but they are short of the inherent support for complex queries that often required by overlying applications. The traditional solutions based on inverted lists for single terms perform are poorly in large- scale distributed settings. Hence,research the shortcoming that document database does not support multiple key when dealing with dynamic set of documents as the primary index,then propose an improved multiterm indexing technique.By storing the inverted lists of combinations of terms,a query- driven mechanism adaptively stores the popular term combinations derived the recent query history. This approach reduces the overall bandwidth consumption,only marginal overhead in terms of additional,but fewrequired storage resources,obviously improving the NoSQL system ＇s capacity and response time.