WD Red Plus vs. Seagate Exos X20

[Peter Jakab]

Hi All,

I am planning upgrade of my old WD Red based NAS. So I would like know what is the pro vs. cons of those drives?
How I see the WD Red Plus WD120EFBX (12TB) similar price as corporate grade SEAGATE EXOS X20 ST18000NM003D (18TB). Both high capacity CMR drives.
I also seen WD offer 1mil MTBF 3years meanwhile Seagate offer 2.5mil 5years.

I was old fan of WD drives (before Red also, yes I am an old FreeNAS member) due have lot of issues with lot of bad Seagate drives. But what about this Seagate corporate drives in TrueNAS CORE system todays? Where is the trick?

 

[sretalla]

Even if you consider the cheaper way to get your hands on some WD "REDs" (or more accurately, whites) is to shuck them from WD Elements/EasyStores external enclosures https://www.truenas.com/community/threads/easystore-drive-shucking-like-a-pro-by-a-pro.89002/

the Exos drives seem to be cheaper per TB and I do see some reports of them working well around here (but not enough time to know how well they will stand up to the 5 years of punishment you will put them through).

 

[ChrisRJ]

The problem with such questions on a forum is that the experience of people is by definition anecdotal. Also, because of totally different environments, workloads, etc. things are not comparable in the least.

With that said here is my personal experience with 8 Exos X16 16 TB drives over the last 3.5 years. It was absolutely abysmal. Tomorrow morning UPS will deliver the 6th or 7th replacement drive. So I am coming close to a 100% failure rate.

I have no idea why this happens. In one case a replacement from Seagate didn't even start up properly. So it cannot only be an initial bad batch of drives. What has to be said to their credit, though, is that the RMA process always worked extremely well.

In total I very much hope that SSD prices for read-intensive models will come down enough so that the next NAS can be flash-only.

This is only my personal experience and perhaps it was the PSU (now replaced as a cautionary measure) that killed the drives. We'll never know. My personal takeaway is that daily backups are crucial and I am happy that things worked in that respect for me. Also, I strongly recommend to plan for double redundancy. So either RAIDZ2 or 3-way mirrors.

 

[Etorix]

How I see the WD Red Plus WD120EFBX (12TB) similar price as corporate grade SEAGATE EXOS X20 ST18000NM003D (18TB).

50% more capacity at the same price? Clear win—not even a contest!
Incidentally, the 5 year warranty gives an indication that Seagate expects to have solved its earlier issues with reliability.

 

[T_PT]

The problem with such questions on a forum is that the experience of people is by definition anecdotal. Also, because of totally different environments, workloads, etc. things are not comparable in the least.

With that said here is my personal experience with 8 Exos X16 16 TB drives over the last 3.5 years. It was absolutely abysmal. Tomorrow morning UPS will deliver the 6th or 7th replacement drive. So I am coming close to a 100% failure rate.

I have no idea why this happens. In one case a replacement from Seagate didn't even start up properly. So it cannot only be an initial bad batch of drives. What has to be said to their credit, though, is that the RMA process always worked extremely well.

In total I very much hope that SSD prices for read-intensive models will come down enough so that the next NAS can be flash-only.

This is only my personal experience and perhaps it was the PSU (now replaced as a cautionary measure) that killed the drives. We'll never know. My personal takeaway is that daily backups are crucial and I am happy that things worked in that respect for me. Also, I strongly recommend to plan for double redundancy. So either RAIDZ2 or 3-way mirrors.

Hi ChrisRJ,

I'm curious about whether the PSU change appears to have helped this issue or whether you're still facing issues with failures? I've been looking at X20 drives, although I'm considering X24 at this point depending on the cost/TB comparison.
I'm suddenly a bit nervous based on your experiences!

 

[ChrisRJ]

Since the PSU change I have not seen issues. BUT: Taking past behavior as an indicator, this does not mean I am out of the woods. There have often been many months between HDDs taken offline.

As to the overall decision: I have made my peace with it for two reasons. Firstly, the RMA process of Seagate worked perfectly for me. Even with paying extra money for a top SLA (which I have not, just the standard warranty for non-OEM drives) I would not expect faster response times and less hassle (in fact there was never any hassle at all).

Secondly, with all those replacements I can be certain that my overall system design works as planned.

 

[beagle]

I haven't had a great experience with my 4TB Exos either. Running a light load on a Dell T630 (mostly file share).

Out of 4 drives, 3 failed over the 5 year warranty window and were RMA'd. The last one just 2 weeks outside.

I have had no issues with HGST and Toshiba corporate grade so far.

 

[T_PT]

Since the PSU change I have not seen issues. BUT: Taking past behavior as an indicator, this does not mean I am out of the woods. There have often been many months between HDDs taken offline.

As to the overall decision: I have made my peace with it for two reasons. Firstly, the RMA process of Seagate worked perfectly for me. Even with paying extra money for a top SLA (which I have not, just the standard warranty for non-OEM drives) I would not expect faster response times and less hassle (in fact there was never any hassle at all).

Secondly, with all those replacements I can be certain that my overall system design works as planned.

Thanks for your response. I was going to use mirrors for the setup I'm planning, but I don't think it will be feasible to go to three way mirrors and as I'm setting up an iSCSI target for VMs, I don't really want to go Z2 or Z3.

Did you have all of your drives in one vdev? was it Z2? Have you had any moments where you thought you could have been in trouble?

 

[T_PT]

I haven't had a great experience with my 4TB Exos either. Running a light load on a Dell T630 (mostly file share).

Out of 4 drives, 3 failed over the 5 year warranty window and were RMA'd. The last one just 2 weeks outside.

I have had no issues with HGST and Toshiba so far.

That's not a good sign either, but I'd hope there's been some level of improvement between the 4TB version and the ones I'm looking at.

Maybe I should at least get some costs on alternatives...

 

[ChrisRJ]

For iSCSI I would always go for SSD mirrors. You can basically forget HDDs, unless we are talking about dozens of them, so that they can add up to a somewhat decent number of IOPS.

My 8 drives are in RAIDZ2. I am not sure what to make from the question about me thinking(!) to be in trouble. Yes, I am mincing words here. But as I like to say "thinking means not-knowing". So in such a context it is never a good idea to rely in feelings.

 

[T_PT]

For iSCSI I would always go for SSD mirrors. You can basically forget HDDs, unless we are talking about dozens of them, so that they can add up to a somewhat decent number of IOPS.

My 8 drives are in RAIDZ2. I am not sure what to make from the question about me thinking(!) to be in trouble. Yes, I am mincing words here. But as I like to say "thinking means not-knowing". So in such a context it is never a good idea to rely in feelings.

I think we could end up going well off-topic for this thread if we open up the conversation about drives... but I am looking at a SLOG, L2ARC and a lot of RAM to mitigate performance issues as much as possible - I did think about a special metadata device too, but I think I'll run out of budget before I can adequately mirror a special vdev. I'm going to be making VMs from four old physical boxes that aren't running anyting too demanding and the server will take more drives later on if that becomes an issue. While the systems aren't demanding, they do hold a lot of data, so cost and capacity is winning against performance.
We do have a lot of data that is read a lot, and that will be presented via SMB or NFS, so I'm hoping a large ARC & L2ARC will hold the hot data and generally avoid hitting the spinning disks. I find it hard to believe that even spinners won't be an improvment over what it's replacing!

Regarding the "could have been in trouble" question - I'd now word it (knowing that you were in RAIDZ2), did you ever get to the point where two drives had failed at the same time and you were waiting for a replacement?
I'd be concerned in that scenario even with a known good backup elsewhere, knowing that I was waiting on a replacement drive and still had a resilver to get through!