All new Registrations are manually reviewed and approved, so a short delay after registration may occur before your account becomes active.
Opinions about RAID 1 and protection against Data Corruption
Hi,
I wanted an opinion from you guys about this article about RAID: https://blog.storagecraft.com/truth-raid-data-corruption/
Mainly about the paragraph:
The goal of RAID, particularly level 1, is creating two equal disks so if one fails physically, your data is still available on the other. Unfortunately, RAID is incapable of determining which of the two disks is bad. Therefore, if no failure is detected, the system assumes everything is accurate and data from the corrupted drive is automatically copied to the second drive.
Is this correct? Is it true that there is no system alarm when one of the drives suffers from bit rot? I still see many hosts offering RAID 1.
So the question is: Is RAID 1 worth having? Has it saved you (a sysadmin/hosting provider) from potential problems?
Thanks.
Comments
I prefer JBOD in personal servers.
Raid 1 or 5 or 10 in production simply because you need to have a combined storage with some protection.
When it rains, it pours. When raid fails, you are going to cry along with your clients.
Repeat after me:
"Raid is not a backup, Raid is to maintain uptime"
>
The other one repeat after me: Backup, backup and again backup
And don't repeat after me: Backup is for pussies. Real men thrive on danger. You live only once. Risk it.
What's a backup?!
I use BTRFS RAID1 so I can have both increased availability (mirroring) as well as data corruption detection (weekly scrubs).
Backup: Back + up ↺
A slang used to refer arse intercourse.
And the last after me : test your backup, test your backup of backup
Yes that is correct and has been demonstrated in tests.
Note however that there are Raid 1 implementations which go beyond merely mirroring. ZFS with its checksums is an example.
Partially, any RAID worth its salt can detect and repair errors that were caused by hardware problems (e.g. bad sectors). Many hardware RAID controllers will scrub to detect these types of errors automatically.
Bit Rot is a completely different problem, and be definition, is not a hardware error. Some RAID 1 implementations, e.g. BTRFS, Ceph, Storage Spaces, etc. can detect and repair Bit Rot automatically. Traditional RAID was never designed with that feature.
Also remember, RAID 1 is not just for data redundancy, but it also improves read performance.
If you're that concerned use an appropriate filesystem, e.g. RAIDZ1, or a btrfs raid if you're a #yolo fellow; RAID alone isn't a protection against bit rot pretty much as it isn't a backup. Whether or not is it worthy to be concerned by bit rot with nowadays' disks is open to discussion and it has been challenged already by some. Chances are that you're wrapping your head for something well beyond your threat model
Real men no need backup 😎🆒
Thank you, everybody, for your views and advice. Much appreciated!
You should be getting larm at os level if 1 disk fails else raid of no use. IT IS better not to have raid but instead a separate backup storage to save data. use disk as it is. btw better go for raid10 but you will end up buying many disks.