摘要:国内的BIM相关技术在近几年发展迅速, 随之高速增长的BIM数据对存储系统的高可用性、负载均衡、横向扩容等功能也提出了更高的要求。本文就BIM行业内某公司需求, 设计并实施了以FastDFS为主HDFS为辅的分布式文件存储系统。该公司主要存储BIM数据中常见的.rfa文件和.rvt文件, 其中, .rfa文件的大小在KB级, 而部分.rvt文件则体量较大。主流分布式存储系统HDFS适合存储大文件, 为节省研发HDFS针对小文件存储算法方面的成本, 本文引入了适合存储中小型文件的轻量级分布式存储系统FastDFS, 以此为主要存储系统, 并部署HDFS, 存储大文件, 为公司之后优化HDFS小文件存储, 迁移全部存储平台至HDFS, 并在此之上建立分布式计算平台打下技术基础。本文在两个存储系统之上布置了配有Keepalived模块的Nginx反向代理服务器, 统一用户存取入口, 在保证系统负载均衡和高可用性能力的同时封装了WebAPI, 依据文件大小, 将BIM数据分别存入FastDFS和HDFS。经测试, 新系统较原系统在存储性能方面有较大提升, 并增添了高可用性、负载均衡、横向扩容等功能。

Abstract: Building information modeling (BIM) related technologies have been widely used in recent years. Dueto itsnon-structural and growing volume, BIM data puts forward higher requirements tostorage system. In this paper, we design and implement a distributed file systemto store BIM data with high availability, load balancing, and horizontalexpansion. The FastDFs and HDFS(Hadoop distributed filesystem) are basically used to store small files and large files respectively. On top of this arrangement, Nginx reverse proxy server with Keepalived module is deployed tounify the user access port, and encapsulate the Web API to manage the BIM data. Preliminary experiment results show that our approach achieves better performance.