Technology Blog

LeoRaid data protection technology

LoongShine Network Copy Link
Abstract: LeoStor is a horizontally scalable NAS system with excellent performance, with dual cluster architecture and intelligent slicing technology. The system supports multiple copies and erasure code mode, and hierarchical storage technology. The system is more stable and responsive.

LeoRaid

LeoRaid is the general name of LeoStor for data protection technology, including multiple copy mode and N+M erasure code mode. When N=1, it can be considered as copy mode, and erasure code N=2, 4, 8, 16, M=1, 2, 3, 4; LeoStor can design a redundant model for a directory separately, instead of only one redundant model for the entire system.

Pattern Strategy Effective space utilization
Copy LeoRaid 1+1 50%
LeoRaid 1+2 33%
LeoRaid 1+3 25%
Erasure code LeoRaid 2+1，N = 2 ，M= 1 66%
LeoRaid 2+2，N = 2 ，M= 2 50%
LeoRaid 4+1，N = 4 ，M= 1 80%
LeoRaid 4+2，N = 4 ，M= 2 66%
LeoRaid 4+3，N = 4 ，M= 3 57%
LeoRaid 4+4，N = 4 ，M= 4 50%
LeoRaid 8+1，N = 8 ，M= 1 88%
LeoRaid 8+2，N = 8 ，M= 2 80%
LeoRaid 8+3，N = 8 ，M= 3 72%
LeoRaid 8+4，N = 8 ，M= 4 66%
LeoRaid 16+1，N = 16 ，M= 1 94%
LeoRaid 16+2，N = 16 ，M= 2 88%
LeoRaid 16+3，N = 16 ，M= 3 84%
LeoRaid 16+4，N = 16 ，M= 4 80%

protection mechanism

LeoStor has a variety of policy protection mechanisms. When the strictest data protection strategy is adopted, it can make data storage more reliable and safer, and reduce the probability of file damage and group extinction caused by hardware failure. The strategy is as follows:

  1. The number of nodes shall not be less than N+M value;
  2. The slice data of a file is distributed on all nodes as much as possible;
  3. The slice data of a file is distributed on all hard disks as much as possible;
  4. The data of a slice group should not be on a node as far as possible, except for temporary recovery;
  5. The data of a slice group should not be on a hard disk as far as possible, except for temporary recovery;

LeoStor allows M nodes to fail at the same time and M hard disks to fail at the same time, that is, N+M: M; When a hard disk failure occurs, the system will restore the data of the failed hard disk to another hard disk at the speed of 1TB/15 minutes. If there is insufficient free space, the system will always be in the state of waiting for restoration. If there is a hard disk failure at this time, the data may be lost. Therefore, the user should 1) reduce the network pressure as much as possible to make the fault data synchronization complete quickly, 2) make the storage space available. It is recommended to reserve at least 3 hard disks or 20% of the spare capacity for the system in case of emergency.

data recovery

For example, the system has seven storage nodes and adopts a LeoRaid 4+2 redundancy model. If the OSS-2 node fails, the system starts the recovery mechanism.

According to the LeoRaid strategy, LeoStor searches for the nodes and hard disks that have not been used by the slice group in the remaining nodes, and then performs the recovery operation of the slice and erasure code data; After a node failure, there are six remaining nodes. When there is another node failure, the system enters the hard disk-based protection mode, namely DiskRaid mode, and the system reads and writes normally.

Users should establish good data backup habits, achieve real-time, scheduled and other multiple strategy backups, and use the LeoStor storage system and other storage devices as heterogeneous backup platforms.

Technology LeoRaid Fault