云环境下多维数据索引结构的研究

发布时间：2018-01-19 14:18

本文关键词： 多维数据索引 skip-webs UB树云索引 SUB-index　出处：《云南大学》2013年硕士论文　论文类型：学位论文

【摘要】：云计算平台作为一种新兴的计算基础设施还需要很多的后续工作,构建云的基础设施最重要的一步就是构建具有良好扩展性的数据存储系统,并且这些数据存储系统必须支持很大的工作量负载。现有比较主流的云存储系统主要包括Google的BigTable, Amazon的Dynamo等。而当前大部分的云存储系统中,多是采用DHT(分布式哈希表)方式来构建数据索引,是一种纯粹的Key-Value模型来组织数据,通过主键高效的获取数据。这类解决方案通常是采用特殊化处理后,基于语义的或者是局部性的映射机制来优化查询的性能,但在多维度查询方面不是很理想。因此本文尝试提出了一种新型的在云环境下的分布式多维数据索引结构SUB-index(skip-webs UB-tree),为解决此类问题提出一些参考。本文尝试提出一种在云环境下多维数据索引结构SUB-index。SUB-index采用UB树这种结构来作为多维数据索引的研究基础,尝试用一种类似线性结构skip-webs架构来构造分布式索引。这样使得SUB-index不仅结合了skip-webs架构中不需要进行平衡调整以代替平衡树中复杂的调整的特性,而且利用UB树考虑到存储结构和磁盘的结构相对应的优势。SUB-index为在云环境下的有效多维数据索引架构提出了新的参考。本文还设计了SUB-index多维数据索引结构的相关查询算法,包括点查询以及范围查询,还有插入和删除的算法。论文最后还对所提出的分布式多维数据索引结构SUB-index的点查询、范围查询、插入更新和可靠性设计了相关的测试和进行了仿真实验。从实验结果可以看出,SUB-index索引结构是高效的,并且在范围查询方面提供了良好的支持。
[Abstract]:Cloud computing platform as a new computing infrastructure needs a lot of follow-up work. The most important step to build cloud infrastructure is to build a data storage system with good scalability. And these data storage systems must support a lot of workload. The existing mainstream cloud storage system mainly includes Google BigTable. In most cloud storage systems, DHT (distributed hash table) is used to build the data index. Is a pure Key-Value model to organize data and get data efficiently through primary keys. This type of solution is typically treated with specialization. Semantic or local mapping mechanisms are used to optimize query performance. But it is not ideal for multi-dimensional query. Therefore, this paper attempts to propose a new distributed multidimensional data index structure SUB-index. in the cloud environment. Skip-webs UB-tree. Some references are put forward to solve this kind of problems. This paper attempts to propose a multi-dimensional data index structure SUB-index.SUB-index in the cloud environment using UB-tree structure as the research basis of multidimensional data index. This paper attempts to construct a distributed index using a linear structure like skip-webs architecture, which not only combines SUB-index with skip-webs architecture, but also does not require balancing adjustment. In place of the properties of complex adjustments in the balance tree. Moreover, using UB-tree to consider the advantages of storage structure and disk structure, the SUB-index provides a new reference for the efficient multidimensional data index architecture in cloud environment. This paper also designs SUB. Correlation query algorithm of index multidimensional data index structure. Including point query and range query, as well as the algorithm of insertion and deletion. Finally, the paper also proposed the distributed multidimensional data index structure SUB-index point query, scope query. Relevant tests and simulation experiments are designed for insert update and reliability. The experimental results show that the SUB-index index structure is efficient. And in the scope of query provides good support.
【学位授予单位】：云南大学
【学位级别】：硕士
【学位授予年份】：2013
【分类号】：TP333

【参考文献】