SSD Endurance

[William Gibson]

The reason I went to SSD back in 2009 was because I had a huge run of failing HDDs. I had 2 hitachi travelstar 500gb HDDs go out and 1 Seagate go out in about 3 months. This was at a time when my macbook never moved off my desk 99.9% of the time. So in 2009 i bought an intel 512 GB SSD and never looked back. I have only had 2 SSDs since then, That OG intel which i sold with that 09 MBP and my current Samsung SSD Pro that is still going strong.

A little bit of hard drive/solid-state drive/general data loss anxiety is a good thing.

Everybody backed up this week? (I haven't yet...)

Ok you got me thinking, Patch. After I started 80 hour work weeks 2 years ago I started paying a guy to warp my tracks. I have just shy of 1k tracks warped by him and have not backed up in 2 years. Cost of warping + cost of actual track = Probaby almost $3000 USD in tracks.

Time to plug in the external RAID drive and get my redundancy on...

[mostapha]

A little bit of hard drive/solid-state drive/general data loss anxiety is a good thing.

Everybody backed up this week? (I haven't yet...)

They're automatic. Yay cron.

[Daniboy]

How can someone argue that a HDD is better than a SSD? How can a freaking rotating disk be better at something? lol

[ImNoDJ]

A little bit of hard drive/solid-state drive/general data loss anxiety is a good thing.

Everybody backed up this week? (I haven't yet...)

I backup daily. You dont fuck around with hard drives. They will bite you in the ass at the worst time, guaranteed.

[ImNoDJ]

How can someone argue that a HDD is better than a SSD? How can a freaking rotating disk be better at something? lol

No one is arguing that, at least not in terms of performance.

Apart from performance, HDD still have their places in large scale storage.

[ImNoDJ]

You're probably right. But, we're not talking about large scale purchasing decisions. We're talking about individuals on a DJ forum. Apart form you, I probably own as many or more drives than 90% of DJTT users.

And, yes, it's my opinion. And theirs. And all it does is show that manufacturer's claims of repeated write performance are conservative. Since those claims are a) as good or better than those for hard drives and b) more than any normal person would need....the remaining "SSDs will eventually die because you can't write to them forever" BS that I see occasionally on here and all over the place elsewhere are fairly conclusively bunk.

I agree with you that the conclusions may be right, but you have to understand that with hard drives, its perfectly probable to get 6 that are not typical of the millions made of that unit.

Thats all Im saying. I dont disagree that there was a lot of FUD about SSDs for a while.

At work we have bought servers that had a cluster of hard drive failures of the same brand, very well regarded, very expensive drives.

We buy other servers that we are waiting to replace when it fails, but even after 4 years nothing has failed yet.

Another user could have bought 1000 of the same drive and not had a single failure. Or they could have had 200 failures.

This actually happens, which is why so many people have superstitions about brand and types of hard drives, when none of it matters in reality.

When talking about failure rates, the only important point is how you backup, not which hard drive will last the longest.

[Daniboy]

No one is arguing that, at least not in terms of performance.

Apart from performance, HDD still have their places in large scale storage.

The first post is about the HDD industry trying to say SSDs wont last because of write or whatever, who would by that cheap talk..? Thats what I am asking

[ImNoDJ]

The first post is about the HDD industry trying to say SSDs wont last because of write or whatever, who would by that cheap talk..? Thats what I am asking

That was Mostaphas original point. There was a lot of fear spread about the reliability of SSDs and a lot of people did buy into it. This experiment was originally designed to address that.

[mostapha]

Apart from performance, HDD still have their places in large scale storage.

Yeah....that's true. Correct me if I'm wrong, but it seems like most of that has more to do with cost/benefit in the presence of other performance bottlenecks. My little 16TB raw NAS was around $1600, IIRC. If I'd done it with SSDs, it would have cost more like $6000. And it would still be limited by the network speed, since 10Gb ethernet stuff is still crazy expensive. And even that is a bit beyond what most people here do.

Also, in a lot of the deployments I've seen, the more modern systems use a decent number of SSDs as cache drives anyway. Mine doesn't yet, but it will next fall when I expand my lab.

When talking about failure rates, the only important point is how you backup, not which hard drive will last the longest.

Agreed. My NAS does not have enough drives from the same lot number or store to take down the whole array if there was some kind of bad run. Other than that, it's just luck, redundancy, and backups.

But we're also not talking about spontaneous failures. I feel like they would have restarted that test if the failure wasn't predictable (notice the graphs on reallocated cells). It's just about that particular piece of FUD, which I still hear.

I think it's mostly salesmen that have no idea what they're talking about but still spout old, out-dated bullshit because it's the only way they know how to sell.