摘要： 海量地形数据的存储与管理是大规模地形实时漫游系统的关键。该文提出一种基于对象存储的分布式并行地形数据服务系统(DPTSS)，采用自治的存储对象存储和管理地形块数据，实现了控制路径和数据路径分离。通过元数据集群提供高效率和高可用的元数据服务，以及基于对象的存储集群实现并行的地形数据块传输服务，提供高吞吐率和高带宽的地形数据服务。对比实验表明，DPTSS在较低的TCO情况下能提供高性能的地形数据服务。

关键词: 基于对象存储, 海量, 分布, 并行, 地形数据

Abstract: The storage and management of massive terrain data is the key of large-scale terrain roaming. This paper proposes a distributed parallel mechanism based on object storage——Distributed Parallel Terrain Service System(DPTSS), which adopts self-rule storage object to store and manage terrain block data and realizes the division of control path and data path. Divided metadata cluster provides high-effective and high-available metadata service, and OSD cluster with parallel terrain data transmission provides high throughout and bandwidth. A prototype system for DPTSS is established. Experimental results show that DPTSS in lower TCO enables good terrain data distributed service.

Key words: Object-Based Storage(OBS), massive, distributed, parallel, terrain data

中图分类号: 

• N945