Application and selection of RAID in monitoring storage

Common level comparison

Recommendations for monitoring scenarios

• 4～8 disks: RAID 5. The usable capacity is high and 1 disk failure is allowed.
• 8 disks or more: RAID 6 is recommended. The more disks there are, the higher the probability that another disk will be damaged during reconstruction.
• Key recordings: RAID 6 or RAID 1 + offsite backup available.
• RAID 0 is not recommended: Although the capacity and performance are optimal, if any disk is damaged, all recordings will be lost.

Reconstruction risk (important)

After a disk in the RAID 5 array fails, the system enters a degraded state. At this time:

• All read requests need to pass verification calculations, and performance drops significantly.
• The reconstruction process needs to read the data of all remaining disks (for example, 8×8TB needs to read about 56TB), which usually takes 10 to 30 hours.
• If another disk fails during the rebuilding process, all data will be lost.
• The URE (Unrecoverable Read Error) probability of large-capacity consumer-grade hard drives makes reconstruction failure not a small probability event.

Practical suggestions

1. Use surveillance-grade or enterprise-grade hard drives to avoid consumer-grade SMR disks (rebuilding is prone to failure).
2. Configure Hot Spare to automatically start rebuilding after a disk failure, shortening the downgrade window.
3. Enable SMART monitoring and alarm and proactively replace it before it is completely damaged.
4. Make dual-machine backup for important projects or save key channel recordings off-site.
5. Rehearse the recovery process regularly to verify that the backup is actually available.
6. When replacing a damaged disk, use a hard disk of the same specification or a larger capacity to avoid mixing disks with different speeds.