为了提高数据密集型计算的计算速度和减少数据传输的通信量,从存储角度出发,设计了一种新的网格存储结构,在设备驱动层实现单一的磁盘空间,采用逻辑卷管理(LVM:Logical Volum e M anager)和独立磁盘冗余阵列(Soft-RAID:Soft RedundantAr...为了提高数据密集型计算的计算速度和减少数据传输的通信量,从存储角度出发,设计了一种新的网格存储结构,在设备驱动层实现单一的磁盘空间,采用逻辑卷管理(LVM:Logical Volum e M anager)和独立磁盘冗余阵列(Soft-RAID:Soft RedundantArray of Independent D isks)的思想,实现了物理磁盘的动态增加和删除,同时实现了容错功能。展开更多
Data replication is a key way to design a disaster tolerance system. This paper presents a replication driver layer-based data replication system on FreeBSD （FRS）. The system is embedded into the replication device ...Data replication is a key way to design a disaster tolerance system. This paper presents a replication driver layer-based data replication system on FreeBSD （FRS）. The system is embedded into the replication device driver layer, does not depend on specific storage devices and logical volume manager, and can achieve replication on data block level. The design considerations and decisions in defining FRS are described in detail.展开更多
文摘为了提高数据密集型计算的计算速度和减少数据传输的通信量,从存储角度出发,设计了一种新的网格存储结构,在设备驱动层实现单一的磁盘空间,采用逻辑卷管理(LVM:Logical Volum e M anager)和独立磁盘冗余阵列(Soft-RAID:Soft RedundantArray of Independent D isks)的思想,实现了物理磁盘的动态增加和删除,同时实现了容错功能。
基金supported by 863 Program under Grant No. 2009AA01A404
文摘Data replication is a key way to design a disaster tolerance system. This paper presents a replication driver layer-based data replication system on FreeBSD （FRS）. The system is embedded into the replication device driver layer, does not depend on specific storage devices and logical volume manager, and can achieve replication on data block level. The design considerations and decisions in defining FRS are described in detail.
