General > General Technical Chat

modern storage is crazy

<< < (6/14) > >>

james_s:
I don't see much advantage there. In my experience most failures are not the platters themselves but the head amplifier or other internal components. I don't think internal RAID would buy you anything, the whole point of RAID is that you can swap out an individual failed drive and rebuild the array.

PlainName:

--- Quote ---Would there be any point in the drive manufacturers offering built in RAID?
--- End quote ---

Only for neophytes. Anyone worth their salt who thinks RAID is going to fix something will, of course, use drives from multiple sources.

eugene:
C'mon guys. I'm not suggesting that it would be as good as other solutions, I'm just wondering if a 2TB internal HD with built-in RAID would somehow protect the data better than a 10TB drive that's 95% empty.

It's a way to make use of those over sized drives in a typical laptop (for example.)

mariush:
No, because RAID is never for data protection. edit : I mean SHOULD never be used for data protection, or as the only data protection method.  There are RAID versions that offer some data protection, in case of a single or multiple drive failures.

A hard drive with internal raid would still have a single point of failure.. let's say a short circuit on the circuit board or over voltage event blows up the eeprom chip that stores the drive geometry info (number of platters, surfaces, tracks, lists of bad sectors, etc) and the controller can no longer access the data.

People/Companies often buy hard drives from multiple locations, or buy drives staggered, just so that they'll get the same model but with different batch numbers, different manufacturing dates.

For example, what if the factory had a small batch of components that were faulty and were all used on hard drives in week 10 of 2021? If you buy 6 drives from the same store to make a raid 5/6 chances are you're gonna get consecutive serial numbers, as the store probably ordered a box of 25-50 drives.  Then, you could have a situation where all six drives fail after a year of 24/7 usage, all within a few days or so from each other.

btw, a bit off topic... .. Seagate even had a firmware bug that cause drives to lock and no longer respond after some amount of time of operation : http://www.datarecoveryspecialists.co.uk/blog/firmware-bug-on-seagate-hard-drive
Had you bought a few drives at the same time, you'd be screwed with all these drives failing at nearly the same time

By making a RAID 5 / 6 or higher with drives with different production dates you reduce the risk of all drives in the raid failing within a few days from each other, or potentially while you spend 10-20 hours to repair the raid by replacing the faulty drive.


If you want data protection, you could have your raid with some parity / spare drive as hot spare, and you'd have some way of periodically (ex at 2am every day or every sunday night) clone the drives to another computer that's in another location (another room so in case of fire you won't have both computers damaged, or at least another power circuit in case you have some power event that blows both computers' power supplies)

Then ideally you'd also copy the critical data to a drive or tapes and put it in a safe, or upload the data to a remote location (a storage server in a datacenter is cheap, 50-100$ a month, you can just encrypt archives with some error correction info (quickpar for example) and upload them to a remote machine.
 

DiTBho:

--- Quote from: mariush on April 28, 2022, 07:34:06 pm ---Seagate even had a firmware bug that cause drives to lock and no longer respond after some amount of time of operation; Had you bought a few drives at the same time, you'd be screwed with all these drives failing at nearly the same time

--- End quote ---

d'oh, I have six of those in my 2009-2010 NAS  :palm:
The article says

--- Quote ---For clients with Seagate Barrcuda 7200.11, DiamondMax 22 and ES.2 SATA hard drives manufactured before 2009, there is a well documented firmware 'bug'.

--- End quote ---
My disks are 7200.12, firmware CC44: I am not sure now if 7200.12/CC44 in the bug-blacklist.

--- Code: ---2010-01-5385, qty=6, Seagate Barracuda 7200.12 SATA 500GB, 005-953, €44,90, €269,40

--- End code ---

Navigation

[0] Message Index

[#] Next page

[*] Previous page

There was an error while thanking
Thanking...
Go to full version
Powered by SMFPacks Advanced Attachments Uploader Mod