Abstract: This paper introduces the spatial query capabilities of MongoDB, which is No-SQL database, and proposes an optimization strategy of the spatial query engine for application scenarios with large concurrent services of massive users. According to the characteristics of little change in content in a short time and high similarity of search results in adjacent locations, this strategy uses a gridstrategy to reduce the number of database operations under the premise of equivalent query results, and uses a memory buffer strategy to reduce disk I/O reading and writing times, thereby improving the concurrency and query speed of the service system. Tests results show that under the same hardware facilities, the optimized search engine efficiency is nearly 50 times higher than that of the original system.

Key words: spatial index, No-SQL, MongoDB, location-based service