WD Gold Vs WD Red for Parity Drive

[irishdab]

Ok so here is a short back story. I just setup my first unraid server. During preclear the WD 8tb Red drive that I had planed on using as my parity drive failed 10 mins in on the first pass. I'm going to RMA it but I've heard you normally get a used older drive back. Not what I want as my parity drive. I was rethinking my needs and since I figure my array will only have 4 tb drives in it I was thinking of buying a new 4tb to use as a parity (I will use my  rma drive for something else.)

So here is my question right now WD gold 4tb drives are pretty much the same price as 4tb red pro and really only a few dollars more then a normally red. That stats on the Gold just seem better then the red (Non Pro). (longer warranty, faster speed,  MTBF is double for the Gold compared to the Red.) Since they are pretty much the same cost is there any reason I should not use Gold drives in my Unraid??

[SSD]

Normally if a drive fails preclear, you can return it to the seller and get a replacement new drive, rather than go to the manufacturer and risk getting a bad refurb. Maybe it was an older drive you repurposed. But if you get a refurb, test it good and return again if it is showing SMART issues.

 

What happened that you think the drive failed. Often what looks like failure is really a cabling or controller problem, and nothing wrong with the disk.

 

Having an.8T parity sets you up to start adding 8T data disks. If you see your array growing significantly, using larger disks is the way to go. Because there is a limit to the number of physical disks your server can hold (and unRaid has a limit even if your server could handle more). Fewer larger disks result in a simpler configuration, less prone to cabling issues, uses less electricity, fewer ports and bays needed, etc. 

 

Considering faster partly vs slower is a reasonable question. Slower rotating drives do slow writes in unRaid normal write mode. And if the parity drive is slow, even if you have faster data disks, parity will always be the gating factor. But on the other side, for many, writes are happening in the background anyway, and even if they are a little slower - who cares. Read performance is unaffected. And the cache / turbo write features speed writes to the server much more than fast parity.

 

Because of how unRaid writes work, rotational speed is important,  but for read performance, platter density is a more important speed consideration for sequential I/O. And bigger disks tend to have denser platters. An 8T 5400 RPM drive might be as fast or faster than a 4T 7200. Of course high platter density and fast rotation combined will be the fastest.

 

As far as drive quality / reliability, you definitely want quality drives. But your most important drives are the ones with data on them. Parity is only called upon should a data drive fail. I can say in my arrays over the years, it has been extremely rare that parity had been relied upon. And when it is, it is only for a few days. Don't be under an illusion that parity is more important than your data drives! And don't be swayed by labeling (gold vs red). Although premium / enterprise drives may be better for constant use, there is no evidence that they are more reliable in light duty mode.

 

If of interest, my drive preference is HGST.  Studies have found them, as a brand, the most reliable. Their 7200 RPM NAS drives are the best you can buy IMO and would make for nice fast parity and data. But they are more expensive than some of the external drives that contain REDS or Seagate drives. I have a mix.

 

One more point, slower drives use less electricity and produce less heat. They also spin slower which might mean longer life due to less mechanical wear and tear. For a large media server, I go with a faster parity but default to slower data drives. Except for a couple disks that I tend to write to more often (eg backups). YMMV

[irishdab]

Thank you for all the information SSD. I'm still very much learning.

The drive that I had that went bad was a new drive. It was built in Jan 2017 I've had it since April sitting on a shelf waiting till I got all the parts  needed. When I tried to preclear it after about 15 mins it started to give me  Current_Pending_Sector errors after a little over an hour the smart had recorded over 7000 of them. I tried to run a short smart test it failed. I then tried to run a long test it also fail. I never had a wd drive fail before. ( I have had other makes fail on me but none so new.) I did wonder about cabling or controller problem so I removed the drive from my Unraid server and put the drive in my windows 10 computer. I launched wd hard drive test software it also fail to be able to run the short test. The long test did run but stopped early and said there were bad sectors on the disk. It gave me the choice to repair but when it tried it failed and told me to contact wd support.  

I have thought about getting HGST but am on the fence. I heard they run hot and loud. But I've never had one before. I have used WD since the Maxtor loud click issue years ago that killed a lot of peoples drives. I was one of them and lost data I could not get back. Now a days I do backups and I am very careful with my data.

 

 

 

[SSD]

2 hours ago, irishdab said:

Thank you for all the information SSD. I'm still very much learning.

The drive that I had that went bad was a new drive. It was built in Jan 2017 I've had it since April sitting on a shelf waiting till I got all the parts  needed. When I tried to preclear it after about 15 mins it started to give me  Current_Pending_Sector errors after a little over an hour the smart had recorded over 7000 of them. I tried to run a short smart test it failed. I then tried to run a long test it also fail. I never had a wd drive fail before. ( I have had other makes fail on me but none so new.) I did wonder about cabling or controller problem so I removed the drive from my Unraid server and put the drive in my windows 10 computer. I launched wd hard drive test software it also fail to be able to run the short test. The long test did run but stopped early and said there were bad sectors on the disk. It gave me the choice to repair but when it tried it failed and told me to contact wd support.  

I have thought about getting HGST but am on the fence. I heard they run hot and loud. But I've never had one before. I have used WD since the Maxtor loud click issue years ago that killed a lot of peoples drives. I was one of them and lost data I could not get back. Now a days I do backups and I am very careful with my data.

 

 

Bad cabling won't cause a pending sector. They are always indicators of a drive problem. Sometimes they mysteriously clear on a new drive if you attempt a second preclear. But sounds like it is truly a bad drive to me.

 

My server sits in my unfinished basement. Unless it screams like a banshee, I don't hear it. But I have never detect HGSTs to be any louder than others.

 

I strongly recommend hot swap drive cases like the CSE-M35T-1B. They have good fans that pull air through the drives and keep them nice and cool. The fans are extremely good - full ball bearing fans. Last forever. But are not the quietest - but some have replaced with quite fans and been happy.

 

If you look at some of the BackBlaze studies they give drive failure rates. HGST normally is the best. Interestingly, HGST is owned by WD. But they have operated mostly autonomously, although recently I've seen some synergies across their drive lines. Both Seagate and WDs have their winners and losers. I hated the autopartking of the WD and didn't buy them for years. Recently bought the 8T REDS that were very cheap, and pretty happy so far.

 

Good luck!

[HellDiverUK]

Golds are the best drives WD make.  They're HGST-based units, so far superior to the Red/Purple/White Helium drives which are lower quality.

 

Though, I really like Seagate's 8TB Ironwolf Pro drives, they're super-quick and quite quiet.