中文摘要：

基于空间数据面向对象存储思想和云存储可扩展架构，将控制信息集中在元数据服务器集群中管理，而实际的空间数据基于对象存储分布到存储设备集群中，实现控制信息路径与数据传输路径的分离，并缓存热点空间数据对象接口，以减少元数据访问次数和降低其服务器负载；基于对象存储设备的并行性和CDMI标准对元数据进行自上而下的功能分层管理，增强动态扩展空间数据管理能力；基于空间数据多雏特性划分元数据访问粒度，进而节约空间数据访问响应时间和提高并行用户数。提出的元数据可伸缩管理方法可适应目前空间数据呈几何增长和服务容量快速扩容的需求。

英文摘要：

Based on spatial data object-based storage and the structural mode of cloud computing storage, the new method of metadata management in this paper kept the control information in the metadata servers and distributed the true spatial data object into object-based storage devices cluster, to realize the separation paths for control information and data translation. And the method caching hot spatial data object interface to lessen the access times to metadata. It lightened the load of metadata servers, so they could expand their ability of spatial object management dynamically. According to the parallel object-based storage devices and CDMI, this paper managed metadata layered functionally, and partitioned the access granularity based on multidimensional characteristic of spatial data, to save the response time and improve the number of concurrent users. The scalable management for metadata can meet the geometric growth of spatial data and their service capacity.