Is WD red even an option after the SMR scandal?

[BaidDSB]

I have been finalizing what i want to buy...

 

Then i remembered the wd red fiasco. But I also read Seagate generally fails more often?

 

 

Which Nas drive should i get?

[porina]

Assume everything can and will fail, and plan for it. That means 1, redundancy and 2, separate backups.

 

Depending on your use case and performance needs, avoid the SMR ones if necessary. I'm running a value/capacity optimised setup with Unraid and I went for Seagate SMR drives simply because they're the cheapest per capacity by a long shot. I have 2 drive redundancy on that, with 10 drives in total (2x Toshiba NAS, 2x Toshiba Enterprise, 6x Seagate SMR) and have a dedicated backup NAS (no redundancy) being populated.

 

If I really cared about performance then the CMR equivalents would add some tens of % cost. Where I am Toshiba and Seagate NAS/Enterprise drives tend to be cheapest per capacity. WD tend to go for more. I've have had many Toshiba HDs in the past without problem. Nowadays my bigger problem is capacity over performance so I don't get them any more.

[OhioYJ]

Pretty much all the manufacturers list the what recording technology the drives use these days, so it's pretty easy to figure out if a drive is CMR or SMR. 

[ewitte]

I've had worse failure with Seagate drives.  

 

1 hour ago, porina said:

Assume everything can and will fail, and plan for it. That means 1, redundancy and 2, separate backups.

 

The problem is even if your data is safe a failure often leaves you with degraded performance, time spent and possibly having to buy something.  Time can be excessive with a SMR drive too.

[porina]

56 minutes ago, ewitte said:

I've had worse failure with Seagate drives.  

Which ones? Reason for asking is, if you asked me 10+ years ago, I would have said the same. They have been trying to improve. I'm giving them a chance today.

 

56 minutes ago, ewitte said:

The problem is even if your data is safe a failure often leaves you with degraded performance, time spent and possibly having to buy something.  Time can be excessive with a SMR drive too.

Unless you can find a drive that never fails, you have to plan for this scenario regardless. Latest Backblaze lifetime annualised failure rates came out at 1.32% overall. My main NAS with 10 drives would be 12.4% chance (about 1 in 8) I might have a drive fail each year. Note it isn't 10x a single drive because statistics. Even if you had a 100 drives you wouldn't be guaranteed to get a failure in a year, as likely it as may be. Anyway, due to my small sample size, luck plays a bigger part than statistics. Bad luck might see one fail at any time.

 

So part 2 of the planning is, what is the recovery steps? I can either wait out RMA or order another replacement to not have to wait as long, while continuing with a single drive failure. What's the rebuild time? I actually did this several times when I reconfigured it. An 8TB SMR drive is about 15 hours to fully write. Even if you got CMR, large capacity drives can take longer than that. 15h risk window is fine for me.

 

All this is not saying the OP should go out and get a SMR drive, I'm only saying if appropriate to the use case, it can be an option. Like if using ZFS, CMR seems to be a requirement over SMR. Maybe OP cares more for performance than I do.

[OddOod]

I think because of the WD Red Scandal, all the major brands are an option. If you'll recall, they all had little mini scandals in the months after that story broke. They were all playing fast and loose. Now it's easy to find the recording technology in the speeds and feeds across the board.

 

1 hour ago, porina said:

Anyway, due to my small sample size, luck plays a bigger part than statistics

This is VERY important to remember. Pretty much no one in the consumer space is going to be in a situation where statistics are playing a bigger role than luck. And I say that having 2 separate 16 drive arrays running. 

[kokosnh]

Any CMR one, should be fine, just remember, to run them in raid 1, or raid 5 ( or equivalent of that ), and still have a backup…