Originally Posted by Jarv0
From what i understand enterprise drives are designed for RAID use. They usually have a built in time limit on error recovery. For example TLER. I have heard that it's always recommended to use enterprise drives for a RAID setup. If the drive doesn't have a built in time limit on error recovery, the drive could halt for a long time trying to recover from errors, and then the RAID controller might treat the drive as failed (and in a RAID 0 that would be a total falure of the array, am I right?)
There are also some consumer drives that are designed for RAID use and as far as I know they should also prevent that from happening. Like WD Red?
My questions are:
1. Does this apply to SSDs? How big of a risk is it to put two consumer SSDs without error recovery time limit, in RAID 0? I'm aware of the risk with RAID 0 itself, but how much "safer" would it generally be to buy enterprise SSDs?
2. Does it apply to Intel Rapid Storage Technology RAIDs? Will RST mark a RAID member as failed if the drive is taking a good time on error recovery?
You can put any hard drive you want in RAID. Personally, I have 2 Seagate ST2000DM001s in RAID 0 in my personal system. Work fine. That is backed up to an external 4Tb drive and also backed up using Backblaze.
The same applies to SSDs. You can really just use whatever you want.
Originally Posted by Lady Fitzgerald
Originally Posted by Raiden
SSDs fail randomly based on terabytes written. Typically they work fine for awhile, and then one day they stop completely. With RAID 0 you lose all data if one of your SSDs die. For some content, that is ok. For example, my steam library is on a RAID 0 array, I can always download games back, and save data is typically sync'd to the cloud. For more important content, I use something called dropbox for more than one copy of my data in the event of catastrophic failure. There are other alternatives as well.
On another server I use RAID 10. 1 set of RAID 0, that is then mirrored to another set of RAID 0. The advantage is 2 disks would need to fail before I'd run into problems. Your second question doesn't quite make sense, one a SSD fails, it's dead. With mechanical drives, you can typically tell because bad sectors will start cropping up before the drive dies completely.
Also.. it's much easier to do data recover on mechanical drives than on SSDs where the data is stored on flash chips instead of magnetic platters.
Anyone who depends on being able to recover data from a failing drive is playing Russian Roulette with their data. You can't depend on getting a warning that a drive is going to fail before it does fail. HDDs can and often do fail suddenly with no warning at all. Professional data recovery is expensive with no guarantee of success. Also, drive failure is not the only way to lose data. Malware, user error, failure of other components in a computer (such as a PSU failing and frying everything inside), power surges and spikes, natural disasters, theft, etc.
The only way to ensure the safety of your data is for it to exist in three places physically separated from each other. Typically, that would be on the computer, on an onsite backup drive and on an offsite backup drives. Backup drives should be kept disconnected from the computer except when updating the backup. Also, RAID of any kind is NOT a backup!
Exactly what I was going to say. Never rely on your array to stay working forever and never rely on a drive to give you warning either. Working in data recovery, I can tell you from experience, many times customers have sent their array in say it just doesn't work. Their controller didn't give a warning. It just stopped.
Fitz, you forgot to mention controller failure. I have had 7 perfectly working RAID drives come in, got RAID parameters but the data became shifted thanks to the RAID controller screwing up. Took a bit of work but it was recovered. However, there was a lot of damage to the file system and data itself.