Hi, I'm trying to pick out 3-4 hard drives for my home server (3 for the server, 1 possibly for desktop) since my current drive 2 have entered "gonna die within a year" and 1 in the "acting funky"
I have about 100GB of data I cannot loose, 1tb of I would rather not loose this, and 500gb of data who cares if its gone.
Right now I'm having a hard time picking out drives since the market seems very fractured and I can't tell which drives are good and which are bad. |
Price is a concern but I would rather pay more now than cheap out and have a drive fail prematurely.
Bonus points: which raid/ZFS/etc drive set up should I use (Server is running Ubuntu 16.04 and its a Z97-A mobo with software raid built in)
but you should get 2 or 3 drives and use offsite or cloud storage if you cannot lose it you could even compress the file or split it up and create 2 free MEGA accounts to store the 100GB
For an array of 3 or 4 TB drives, nothing less than a raid 6 should be used if you cannot lose data. ZFS is definitely your go to option to stave off bit rot and read / errors. I would recommend 4 or more 3/4TB drives, in Raid Z2 (aka raid 6 in ZFS).
The Western Digital/HGST drives are very reliable.
he RED drive have TLER :
TLER (Time Limited Error Recovery) is a very important feature if you will be using the drive(s) in RAID since it limits the amount of time a hard drive can spend trying to recover from an error. Normally, if a data error is detected the hard drive will attempt to recover the data and repair the error. Depending on the nature and severity of the error, this can take anywhere from a few milliseconds to a couple of minutes. Normally, this isn't a big deal - it just results in the data not being available for a bit longer than normal. However, RAID controllers only allow for a short amount of recovery time (usually about 7-14 seconds) before the controller assumes that the drive is having problems, drops the drive from the array, and marks the array as degraded.
What TLER does is limit the amount of time the hard drive can spend trying to repair an error before giving up. Since many types of RAID have built-in error correcting, it is preferable to let the RAID itself repair the error than to let the hard drive drop and degrade the RAID array. While TLER is absolutely great if the drive is used in a RAID array, it is not useful (and could even be a negative) if the drive is simply a stand-alone drive. Luckily, WD has provided us with proprietary tools that allow us disable TLER on systems that do not have RAID arrays.
@AFellowGamer I have heard their numbers are somestimes skewed, but genrally say WD or HGST are the go tos, which I have a brand bias.
@thecaveman I thought Raid 6 you need at least 5 drives? 3 for data, 2 for parity? Also not sure motherboard supports it I am leaning towards ZFS, snce it would give me great experience and I can tell employers I do smart things, but from your post its suggesting I need 4 drives in the server (Which can be done I suppose if I forgo upgrading desktp storage)
I have also been told Raid 1 + JBODs would be a valid option
@Miguel_Sensacion I see, so the Reds have a hardware feature that is almost required for doing raid successfully, I ask since newegg has WD blues for $110, and the Reds are $140, which for that price I could get 1 more Drive - but then again why are they so much cheaper compared to Seagates 3TBs and the rest etc
I actually have a 2TB {1TB+1TB} RAID0, running for a year in my rig partition may different ways. I also have a 3TB WD Black in my rig for storage. Not including the other 6 HDD's I have outgrown over the years that still work. Western Digital is a trusted brand for me and statistically {including the aquisition of HGST} is industry reliable. Seagate is a roll of the dice, You get good one and you get bad ones...
Please take in mind that a back up of important data is crucial, if that means buying a 1yr Mega account or Amazon Cloud for those important files or even buying a USB 1TB HDD...
I currently use Seagate Constellation (Enterprise) drives in my server, and they are really fast for HDD standards, plus come with a 5 year warranty, but you pay a premium for that. I've heard that Hitachi's Ultrastar drives are also very fast and reliable, but I cannot vouch for that (you can most likely find benchmarks online for them). If you like WD products, you should probably go for their WD Red PRO or WD Data Center drives, and not the standard WD Red drive, because the standard Red Drive's are aimed more toward basic consumer tasks (slower speed, less performance), and wouldn't be as reliable as all the other drives.
Edit: No matter what you get, make sure the drive has a minimum of a 5 year warranty, if you really value the data. Normally, if the drive has only the standard 2/3 year warranty, it isn't meant to be left on 24/7, with constant reads/writes, and those tend to die pretty quick in the environment.
Have 8xWDC WD3000FYYZ. After 3 years of 24x7 in RAID5, one of them started collecting reallocated sectors counter (easily RMA'd because of the 5yr warranty), another one has maybe 20 of the same (but doesn't increase, so I didn't bother to RMA it yet), 6 others are happy and error-free.