| Currently, Spatial information applications encountered a serious bottleneck, on the one hand,it has accumulated a huge amount of spatial data, and also continue to invest heavily in production data, on the other hand, a large number of spatial data can not be integrated treatment and effective use. In order for efficient storage and management of data, relational database technology came into being. But in the web2.0era, in the face of large-scale and high-concurrency applications, relational databases seem to weakness all show, the reason is relational database consider too many features, such as transaction paradigm constraints, greatly influenced the database performance and scalability. The NoSQL databases abandoned relational database transaction consistency and constraints of the paradigm, in large measure to solve the many challenges faced by the relational database.This study is based on the Redis vector data organization. As a key-value NoSQL database, Redis still has high concurrency, high performance, rich data structures, suitable for the organization and management of complex vector data. Vector database management, in accordance with the quaternary structure of the vector database, spatial data sets, layers, elements of the vector data-level organizations, specifically design the layers of meta-data structure and selection of an efficient Redis data structure metadata information and vector data (geometry and attributes) for storage. Spatial index, in-depth study Redis-based spatial index organizations and query mechanism, and a grid index, for example, based on Redis grid index expression, the storage structure and query; traditional grid index shortcomings, based on small-angle-coded grid index:first small angle encoding vector data, and then use Redis ordered set data structure stored grid cell index information table, on this basis, an improved grid index method-small-angle coding algorithm fast indexing query. By small angle encoding ideological construct a grid cell index information table automatically divided into two parts, repeat region and non-repetitive region only need to query the index information for the repeat region using a simple comparison can be done in O (N) time re-check. No maintenance due to re-check to find a tree like data structure, which reduces the space complexity and improve the efficiency of the index of the vector data grid.Finally, the paper designed and implemented based on the Redis vector data engine prototype system, and the use of large quantities of vector data for comparative experiments and performance analysis, and experimental results verify the basic idea of this paper. |
