›› 2014, Vol. 16 ›› Issue (2): 165-172.doi: 10.3724/SP.J.1047.2014.00165

• ARTICLES • Previous Articles     Next Articles

Research of Lightweight Vector Geographic Data Management Based on Main Memory Database Redis

ZHU Jin, HU Bin, SHAO Hua, LUO Qing, JIANG Nan, ZHANG Jingyun   

  1. Key Laboratory of Virtual Geographic Environment, Ministry of Education, Nanjing Normal University, Nanjing 210023, China
  • Received:2013-08-09 Revised:2013-11-11 Online:2014-03-10 Published:2014-03-10
  • Contact: 10.3724/SP.J.1047.2014.00165 E-mail:hb_hubin@126.com

Abstract:

Effective organization and management of vector geographic data is one of the key parts for spatial database application. The traditional vector geographic data service is usually based on magnetic disks and relational databases like Oracle Spatial. With the rapid development of wireless communication and mobile web technology, the performance of current vector geographic data service is declining dramatically under multi-user concurrent access, and can't meet the requirements of high performance and high concurrency. In order to improve the performance of vector geographic data service under multi-user concurrent access, we proposed a novel management approach of lightweight vector geographic data based on main memory database Redis. Redis is a main-memory lightweight key/value store. Its I/O performance is much better than traditional disk-based databases like Oracle and MS SQL server. At first, we analyzed Redis' key-value data model and data structure. Subsequently we designed a four level hierarchy organization structure of vector geographic database. We stored vector geographic data and its metadata based on Redis' plentiful data structures. Then, taking the grid spatial index as an example, we designed the storage method of spatial index and spatial query processing flow for Redis based on the hierarchy organization structure of vector geographic database. Our experimental results confirmed that compared to traditional relational spatial database-Oracle Spatial, our main memory style vector geographic data management approach greatly improves spatial query responding speed and its concurrent performance is excellent. The proposed approach can be used as a front end high performance cache of large spatial database or a high performance spatial indexes database.

Key words: vector geographic data, main memory database, spatial index, key value database, Redis