Western Digital is releasing Dual-Actuator HDD with performance close to SSDs

[Doobeedoo]

15 hours ago, Crunchy Dragon said:

I remember when WD made 10 and 15,000 RPM hard drives.

 

Be interesting to see a combo of the two.

Yeah also Seagate Cheetah and I think there were even 20K RPM drives. Imagine that with multi actuator for each plate and being hybrid SSHDD too hah.

[Beerzerker]

6 hours ago, LAwLz said:

I don't think using HDD over SSD because "it usually doesn't fail outright" is a very good idea. Yes it's true that SSDs are more likely to fail "out of the blue" than HDDs, but the frequency of which they fail is so incredibly low compared to HDDs that it doesn't really matter.

 

If you want some hard numbers, a Western Digital HDD is around 650% more likely to break within the first year compared to a Samsung SSD. If we are talking consumer drives. At least according to the latest numbers I could find (which was sadly 2017, but I doubt much has changed).

 

It's not enough to just look at "what happens if a drive fails" to do risk assessment. You also have to factor in how likely it is for a particular drive to fail, and in that regard SSDs are far and beyond more reliable.

ANY drive can and will fail sooner or later so that statement is kinda off target.

The point is about rescuing data BEFORE the drive dies regardless of drive type since when your data is inaccessible - It's gone and that's it.
Yes, SSD's do fail as well with the same frequency of failures too by the numbers vs what's out there (Percentage of SSD's and HDD's currently in use), it's just that SSD's haven't been around as long as HDD's have so there is a disparity in the failure count in play by said numbers.

Also know this is based on what's reported - I can promise you alot of these drives regardless of type fail and are simply replaced with no reporting of it done.

We're at a point right now many HDD's are at or close to the end of their life cycle and that's why you have the higher number of failures reported right now, that will fall off in a few years once the majority of these older HDD's die off, then it will naturally shift over to SSD's having more failures.

Even with that I'd rather have a HDD for having the best chance of not losing data and at least a chance of some warning it's gonna die so I can take action and have a shot at not losing stuff I cannot replace.

[Bitter]

I was asking myself about dual head drives like 20yrs ago when my friend bought a Quantum Fireball, seemed like a no-brainer!

Edit: I forgot a whole decade, 20yrs not 10!!! Pentium 4 and Rambus was the new hotness, literally.

[Beerzerker]

5 minutes ago, Bitter said:

I was asking myself about dual head drives like 10yrs ago when my friend bought a Quantum Fireball, seemed like a no-brainer!

Got a Fireball here myself - Slower than cold dog poo but it still works with that Oh-So classic "Noise" all HDD's back in the day used to make.

[TetraSky]

Interesting. Faster speeds than regular HDD, while also having larger capacity than SATA SSD.

Could be good for a mass storage that you need somewhat fast access to, like most games.

 

But at the end of the day, it's still a lot of moving parts that have a greater chance of failure compared to SSD and the random reads/writes will still suck balls compared to a SSD that can access all of its files at the same speed.

[da na]

17 hours ago, Crunchy Dragon said:

I remember when WD made 10 and 15,000 RPM hard drives.

 

Be interesting to see a combo of the two.

High RPM drives are still on the market, but more on the server side - WD did try to break into the prosumer market with the Raptor/Velociraptor line of SATA 10K drives. It's not feasible IMO to have dual actuator high RPM unless price can go way down - too expensive and hot for consumers, too slow for the price compared to SAS SSDs for servers.

[LAwLz]

2 hours ago, Beerzerker said:

ANY drive can and will fail sooner or later so that statement is kinda off target.

The point is about rescuing data BEFORE the drive dies regardless of drive type since when your data is inaccessible - It's gone and that's it.
Yes, SSD's do fail as well with the same frequency of failures too by the numbers vs what's out there (Percentage of SSD's and HDD's currently in use), it's just that SSD's haven't been around as long as HDD's have so there is a disparity in the failure count in play by said numbers.

Also know this is based on what's reported - I can promise you alot of these drives regardless of type fail and are simply replaced with no reporting of it done.

We're at a point right now many HDD's are at or close to the end of their life cycle and that's why you have the higher number of failures reported right now, that will fall off in a few years once the majority of these older HDD's die off, then it will naturally shift over to SSD's having more failures.

Even with that I'd rather have a HDD for having the best chance of not losing data and at least a chance of some warning it's gonna die so I can take action and have a shot at not losing stuff I cannot replace.

The numbers I quoted were based on failure rates within 1 year, so it's an equal playing field. 

 

Facts are that a HDD is far more likely to prematurely die than an SSD. Of course both drives will probably die if you use them for let's say 40 years, but that's kind of irrelevant. What's important is if they die prematurely after let's say a year or two. For that, SSDs are dramatically more reliable. 

 

If you buy a WD HDD and use it for 5 years, you are far more likely to have the drive die compared to if you had bought a Samsung SSD. That's the reality. And it's not a small difference either. It's a very big one.

You could buy 5 Samsung SSDs and the likelyhood of a single one of them breaking is as big as the likelihood of a sole WD HDD breaking if all of them were in service for let's say 1 year. 

 

HDDs are not as reliable as SSDs. They fail in a different way than an SSD, and the way they fail has some benefits, but the rate of which they fail is way higher. 

[wanderingfool2]

8 hours ago, LAwLz said:

I don't think using HDD over SSD because "it usually doesn't fail outright" is a very good idea. Yes it's true that SSDs are more likely to fail "out of the blue" than HDDs, but the frequency of which they fail is so incredibly low compared to HDDs that it doesn't really matter.

 

If you want some hard numbers, a Western Digital HDD is around 650% more likely to break within the first year compared to a Samsung SSD. If we are talking consumer drives. At least according to the latest numbers I could find (which was sadly 2017, but I doubt much has changed).

 

It's not enough to just look at "what happens if a drive fails" to do risk assessment. You also have to factor in how likely it is for a particular drive to fail, and in that regard SSDs are far and beyond more reliable.

It's a valid reason though when comparing HDD's to SSD's.  Unlike the people who say the reasons for a HDD's is only the cost/storage capacity, this just gives more reasons why HDD's can be better than SSD's.

 

Lots of consumer level people don't keep spare drives/ssd's around to swap out when their device crashes.  That's where the "out of the blue" can really come back to make life a pain...or when you just need it to work and it decides that it's going to stop then.  With a harddrive, the lead up to failure is often quite evident (whether by noise, smart warnings, Windows warnings, slowing down of accessing files, randomly having files unaccessible unless you try again).  At least with HDD's you can usually limp along (At my work, people still stores things locally despite being told not to...human nature I guess, when their HDD crashes at least some of the work usually can be recovered which is not the case for the SSD users)

 

The 6.5x number as well is also a highly misleading statistic unless you present the failure rate of the SSD/HDD itself.  In an extreme example, imagine failure rates were 0.00001% for SSD's and HDD's were 0.000065% failure rates each year.  You could claim the "650% more likely", but the issue becomes "does it matter" because in that extreme scenario you would have 1 in 10 mill drives fail vs 6 in 10 mill.

 

Backblaze, which I do have issues with how they typically present/generate data, at the 14.3 month mark failure rates were 1.05% SSD, 1.38% HDD's (Keep in mind they shuck drives).  So yea, if I buy 10 SSD's and 10 HDD's, with the SSD's I would have about a 10% shot at having one or more of them broken within a year vs 13% with HDD's.  I personally have a stack of SSD's that have failed in my house, and it's size is quite a bit bigger than the HDD's (which still technically work, but have effectively failed...although I have thrashed a lot of SSD's by mistake). 

 

Another way to look at it, if you bough 1 SSD vs 1 HDD at the 5 year runtime mark, you expect 4% chance an SSD dies, vs 8.6% for an HDD (based on the backblaze data...then again backblaze shucks drives, keeps them running 24/7 where the physical components like the motor are more likely to fail and they remove the drives when smart data suggests they might fail...SSD's they just let them die since they really didn't get warnings)

 

[NumLock21]

20 hours ago, Crunchy Dragon said:

I remember when WD made 10 and 15,000 RPM hard drives.

 

Be interesting to see a combo of the two.

WD only made 10K RPM drives called the Raptor and later on Velociraptor, both of them uses standard SATA. Seagate was the one that made 15K RPM drives called the Cheetah. Back then it ran on the SCSI interface and these days it runs on SAS, which is the successor to SCSI.

[LAwLz]

2 hours ago, wanderingfool2 said:

The 6.5x number as well is also a highly misleading statistic unless you present the failure rate of the SSD/HDD itself.  In an extreme example, imagine failure rates were 0.00001% for SSD's and HDD's were 0.000065% failure rates each year.  You could claim the "650% more likely", but the issue becomes "does it matter" because in that extreme scenario you would have 1 in 10 mill drives fail vs 6 in 10 mill.

The statistics I quoted stated that Samsung SSDs bought by consumers had a failure rate of 0,17% in the first year.

Wester Digital HDDs had a failure rate of 1,26% within the first year.

That is a big and meaningful difference since typically, if a drive survives more than a year it will keep working for a long time, when talking about consumer drives in desktops (which is what the statistics I am quoting talks about, not Blazeback which is irrelevant for consumers).

 

How the drives dies can be a benefit for HDDs, but it's such a small benefit that when talking about consumer drives I strongly believe the added risk of the drive dying at all greatly outweighs that benefit.

I mean, even your own stats you bring up (which is from BackBlaze and I absolutely hate when people use those because their numbers are meaningless garbage) puts the HDD at twice the failure rate, but please note that BackBlaze should not be used as a source for information because their info is completely meaningless as we discuss in this thread in 2021.

[wanderingfool2]

  

2 hours ago, LAwLz said:

The statistics I quoted stated that Samsung SSDs bought by consumers had a failure rate of 0,17% in the first year.

Wester Digital HDDs had a failure rate of 1,26% within the first year.

That is a big and meaningful difference since typically, if a drive survives more than a year it will keep working for a long time, when talking about consumer drives in desktops (which is what the statistics I am quoting talks about, not Blazeback which is irrelevant for consumers).

So you chose the lowest return rate SSD listed and then chose the highest return rate HDD that the articles mentioned.  You can tell how bad of a sample size they have is as well, since they have Corsair SSD at 1.67% prior and Seagate HDD at 0.8% prior.

 

So yea, your numbers are highly cherry-picked; at least google translate says it's return rate not failure rate.  If you even clicked on the conclusion page you would have seen that during the periods P3 - P9 SSD's were quite a bit higher return rates.

 

Even if were were to ignore that, looking at the previous 2 periods P15, P16; you have 0.86% vs 0.38% and 1.04% vs 0.28%.  That doesn't really tell failure rates though, just return rates.  If you drop a HDD in shipping it's more likely to be DOA as well.

2 hours ago, LAwLz said:

I mean, even your own stats you bring up (which is from BackBlaze and I absolutely hate when people use those because their numbers are meaningless garbage) puts the HDD at twice the failure rate, but please note that BackBlaze should not be used as a source for information because their info is completely meaningless as we discuss in this thread in 2021.

You claim to hate it when people bring up BackBlaze, but the reality is they have at least numbers that you can check against (smart data, etc).  They publish their datapoints, which really does mean you can at least gain insight from it (Because realistically it's like a worst case scenario that BackBlaze is putting their drives through, higher read/writes, 24/7 on time, likely higher vibration etc).  Couple that with the fact that their SSD's appear to fail due to controller fault then yea, it can be used for data.

 

Compare that to your source, where you cherry picked numbers to make it look the best for you.  You failed to even realize that they are talking about return rates, not failure rates.  If I ordered a box of 10 HDD's and the box came damaged I would return them (doesn't mean they failed).

 

Again, you are missing the point as well.  It doesn't matter even if it's double the failure rate (which btw, it's not), the fact is the failure rate is low enough that it's less of a meaningful difference.  Again, 5 years 1 HDD and 1 SSD, 13% vs 10% that the drive would have failed.

 

Also, that stats I brought up in this thread 1.05% vs 1.38% is in no way double.  Double the failure rate would be 2.10% for HDD, it's 1.38% which makes it 1.31x

 

As was discussed in the previous thread as well (where you incorrectly read it as well), you can gain information from it so please stop bashing it as meaningless garbage unless you actually bother to think about the data that you are viewing instead of coming to meaningless conclusions about the data without any idea what is actually in front of your eyes.

[Zodiark1593]

5 hours ago, wanderingfool2 said:

It's a valid reason though when comparing HDD's to SSD's.  Unlike the people who say the reasons for a HDD's is only the cost/storage capacity, this just gives more reasons why HDD's can be better than SSD's.

 

 

 

Is this a valid reason when having to rely upon the failure mode to protect your data, tends to mean you’ve done something quite badly wrong. (In this case, lack of backups)

[leadeater]

@wanderingfool2SSDs I think used to be a lot better reliability wise, sadly I think they have become victims of the race to bottom dollar. Also why is it so common for SSD controllers to fail yet HDDs controllers, yea they have them, basically never do. I find that quite annoying. It may be that due to the I/O rates the SSD controllers have higher workloads and run hotter but like, design for that so it's not a problem.

 

I don't know if its true or not but I swear SSDs have gotten worse, or at least low quality ones more popular due to the shift away from HDDs and trying to get "enough" SSD capacity to not need an HDD. One of the key benefits of SSDs was that reliability potential so sacrificing that so save a little money seems wrong to me.

[Beerzerker]

5 hours ago, LAwLz said:

The numbers I quoted were based on failure rates within 1 year, so it's an equal playing field. 

 

Facts are that a HDD is far more likely to prematurely die than an SSD. Of course both drives will probably die if you use them for let's say 40 years, but that's kind of irrelevant. What's important is if they die prematurely after let's say a year or two. For that, SSDs are dramatically more reliable. 

The thing about how many HDD's are still out there coming up to their expected lifespan today is relevant like any other thing that has an expected "Lifespan", which all things do.
Once that time has passed SSD's will catch up in terms of failures by the numbers.

5 hours ago, LAwLz said:

If you buy a WD HDD and use it for 5 years, you are far more likely to have the drive die compared to if you had bought a Samsung SSD. That's the reality. And it's not a small difference either. It's a very big one.

You could buy 5 Samsung SSDs and the likelyhood of a single one of them breaking is as big as the likelihood of a sole WD HDD breaking if all of them were in service for let's say 1 year. 

I've had more SSD's die on me in a single year (4) than I've ever had any HDD's die off period, and that's from about 20 years of use across the board with all of them.

Of those SDD's, three were less than 6 months old from being new out of the box so don't be suprised about my thoughts on it. 
They died as said, no warning or anything - Just dead and that was it, whatever was on them was still there with no way to retrieve any of it.

5 hours ago, LAwLz said:

 

HDDs are not as reliable as SSDs. They fail in a different way than an SSD, and the way they fail has some benefits, but the rate of which they fail is way higher. 

I did have an older WD 30GB HDD to die on me and that was after 10+ years of daily use and in that instance it did start making noises and acting weird before it finally died.
I was able to get all I needed from it before it finally went out and that to me is a very relevant thing about it.

To me the fact something will fail/die sonner or later is reality - Just how it is and there is nothing any of us can do about it. 
My thing is the potential ability to not lose anything except for a drive itself when that time comes, which it will with whatever you go with.
 

[wanderingfool2]

10 minutes ago, Zodiark1593 said:

Is this a valid reason when having to rely upon the failure mode to protect your data, tends to mean you’ve done something quite badly wrong. (In this case, lack of backups)

Backups aren't always realistic to happen always, also having spare drives for when things fail isn't always realistic.  Let me also ask it like this, is your data currently backed up to this precise moment?  If it were to completely crash, what would you be missing compared to the last backup.

 

Also, the fact it gives a warning is completely a valid reason because again, you don't have spare drives hanging around...at least if you realize something is wrong you can replace it before it breaks.  Hard drives from my experience rarely just completely break, most of the time you can have a few weeks to months before it finally breaks.

 

15 minutes ago, leadeater said:

SSDs I think used to be a lot better reliability wise, sadly I think they have become victims of the race to bottom dollar. Also why is it so common for SSD controllers to fail yet HDDs controllers, yea they have them, basically never do. I find that quite annoying. It may be that due to the I/O rates the SSD controllers have higher workloads and run hotter but like, design for that so it's not a problem.

Yea, I always assumed the controllers failed because it was just always running at higher temps, clock speeds and because the controller needs to generally do more than what a HDD one is doing...so figured it was a complexity thing that made it more prone to failure.  I don't know though.

 

I do think you are right though, more push to capacity by having more MLC reducing the amount the NAND can handle mixed in with the "supply shortages" so some of the makers are using more inferior chips.  Like honestly, aside from Samsung I don't know who to trust for SSD's and even then they had a firmware bug in the new 960's that was causing a large amount of wear to occur.

[Quackers101]

considering there was a recent update for the 980 samsung models and other issues. so SSDs can have issues with price, etc.

Also when there was defective products being sold? More SSDs that own some of the market, more leeway they have, at least some of the prices have gone down over the years.

 

Also that SSDs can only write so much, compared to HDDs.

[leadeater]

14 minutes ago, Beerzerker said:

Once that time has passed SSD's will catch up in terms of failures by the numbers.

Failures by numbers is literally pointless, you know that right? Probability is what matters. If you personally only buy 1 SSD and 1 HDD then you personally are effected by the probability of failure and that is or was lower on SSDs.

 

The number of enterprise SSDs we've had fail is zero, the number of enterprise HDDs we've had fail is in to the thousands. We have a lot of both, more HDDs. They have the same lifecycle.

 

Mathematically HDDs have a higher failure rate given the equivalent class of product. If you buy trash then you only get what you pay for.

 

It is of no surprise that cheap low quality HDDs fail at higher rates than more expensive ones, as is the same for SSDs. If people actually want a reliable SSD then buy one.

 

I actually find it dumb that people go out of their way to buy a WD Red or Seagate Ironwolf then turn around a buy near as much the cheapest SSD they can get and then complain it fails.

[leadeater]

16 minutes ago, Quackers101 said:

Also that SSDs can only write so much, compared to HDDs.

Write wear is so damn high this is a non issue bar maybe the worst SSDs possible. When I've brought cheap SSDs on purpose and they failed it's always been the controller. Thing is I brought them expecting them to fail.

[Beerzerker]

47 minutes ago, leadeater said:

Failures by numbers is literally pointless, you know that right? Probability is what matters. If you personally only buy 1 SSD and 1 HDD then you personally are effected by the probability of failure and that is or was lower on SSDs.

Actually it does have a point because numbers add up to a percentage of failures based on what's out there, what was sold within a given time frame and so on, which goes to my next point.

47 minutes ago, leadeater said:

The number of enterprise SSDs we've had fail is zero, the number of enterprise HDDs we've had fail is in to the thousands. We have a lot of both, more HDDs. They have the same lifecycle.

I'm not suprised you've had more HDD's die due to the fact you have more of them to die in the first place (If I understood you correctly), making the point about numbers itself relevant if so.
If you have more of them and they are in use, it's only natural you'll have more failures of them.

If you had more to die with similar numbers or even less in use that's a thing too but even with that a percentage still applies.

47 minutes ago, leadeater said:

It is of no surprise that cheap low quality HDDs fail at higher rates than more expensive ones, as is the same for SSDs. If people actually want a reliable SSD then buy one.

No arguement here, anytime you buy low tier stuff, expect low tier results because that's probrably what you'll get.

47 minutes ago, leadeater said:

I actually find it dumb that people go out of their way to buy a WD Red or Seagate Ironwolf then turn around a buy near as much the cheapest SSD they can get and then complain it fails.

Again that refers to low tier vs a higher tiered product.
I never buy the cheap stuff because what I get I want to last.

I was always catching flak over what I was buying for boards for example; I never bought many lower tiered models because I knew they woudn't hold up to what I'd be doing (Ln2 benching) so I always bought the Sabertooths and Crosshairs, Apex and Impact boards instead of the lower models available.

It is true, you can abuse hardware no matter what it is and have it to fold, I bought what I did because it was made and suited for what I was doing so naturally I got those - In fact I'm posting with one of them right now (Sabertooth 2.0) that was ran under any condition you can name for over 7 years and it's still working.
It does have an issue with it always thinking it's had a new CPU installed everytime I boot it (That's why I "Retired" it from benching last year) but that's not a problem to get around and aside from that, it still works like it should - BTW it was bought as an "Open Box" item instead of brand new because of what I was going to use it for.
To that end, I believe I've gotten my money's worth and then some from it.

In contrast to that point, if I had tried all that with a lesser board model it woudn't have lasted as long or performed as well either, the same can be said about drives or about anything else you can put a name to.

One other point to make is even if you buy "The good stuff", that is no guarantee of how long it's going to last or how well it's going to work - Even the best have their fair share of lemons.

[leadeater]

3 hours ago, Beerzerker said:

I'm not suprised you've had more HDD's die due to the fact you have more of them to die in the first place (If I understood you correctly), making the point about numbers itself relevant if so.
If you have more of them and they are in use, it's only natural you'll have more failures of them.

Correct but it doesn't change the mathematics on the probability of an HDD failure. You're mistaking probability of experiencing a failure with probability of a failure.

 

Having more HDDs does not make any one HDD more likely to fail. Having more HDDs makes more likely you will have an HDD fail. Not the same thing.

 

If you buy one HDD and one SSD the probability of failure is higher for an HDD, there being literally millions/billions of HDDs in existence and usage right now doesn't change this for you. It only gets messy when you have to factor in the quality of the part in the first place, since I do expect both a low end SSD and a low end HDD to have higher failure rates than high quality ones.

 

Which is why no by the numbers does not matter, if you are just going to totals and nothing else then this is not a relevant data point that is informative of anything. The problem with that you said was your brought in lifecycle, which is why I specifically mentioned the lifecycle was the same. That comes back to the issue that you were trying to point to that more HDDs have been sold in the past than SSDs so the total number being seen is higher, while that is true that still doesn't change the probability of failure of an HDD or an SSD for that matter.

 

The fact the SSDs will catch up in terms of numbers of failures doesn't change the reliability of SSDs or HDDs. I'm not going to buy an HDD is 5 years time because more SSDs are failing in total, that doesn't mean my HDD I just purchased is less likely to fail. Hence by the numbers is pointless.

[williamcll]

They could probably solve the random write speed issue with a SO-DIMM slot on the drive.

[starsmine]

25 minutes ago, williamcll said:

They could probably solve the random write speed issue with a SO-DIMM slot on the drive.

primocache the intel optane on firesale. 

[williamcll]

9 minutes ago, starsmine said:

primocache the intel optane on firesale. 

no this takes up the RAM slots on my motherboard

[starsmine]

7 minutes ago, williamcll said:

no this takes up the RAM slots on my motherboard

No, not optane memory, optane

... This is is part of why optane failed. 

[Beerzerker]

3 hours ago, leadeater said:

Correct but it doesn't change the mathematics on the probability of an HDD failure. You're mistaking probability of experiencing a failure with probability of a failure.

Which is really saying the same in a different way with how you put it.
Experiencing a failure is still a failure, making them one in the same.

3 hours ago, leadeater said:

Having more HDDs does not make any one HDD more likely to fail.

 

Not what I was getting at.

3 hours ago, leadeater said:

Having more HDDs makes more likely you will have an HDD fail. Not the same thing.

Having more does not make any one example more at risk to fail, it's the probrability of having a failure in the first place that goes up since each drive is another chance of a failure occuring, which was one of my points so we agree here.

 

3 hours ago, leadeater said:

If you buy one HDD and one SSD the probability of failure is higher for an HDD, there being literally millions/billions of HDDs in existence and usage right now doesn't change this for you. It only gets messy when you have to factor in the quality of the part in the first place, since I do expect both a low end SSD and a low end HDD to have higher failure rates than high quality ones.

Again we agree, goes back to you get what you pay for.

3 hours ago, leadeater said:

Which is why no by the numbers does not matter, if you are just going to totals and nothing else then this is not a relevant data point that is informative of anything. The problem with that you said was your brought in lifecycle, which is why I specifically mentioned the lifecycle was the same. That comes back to the issue that you were trying to point to that more HDDs have been sold in the past than SSDs so the total number being seen is higher, while that is true that still doesn't change the probability of failure of an HDD or an SSD for that matter.

Nope - My point was right now what's coming to the end of it's life cycle concerning HDD's vs SSD's, it's HDD's that's getting there by the numbers vs what you'd expect to see with SSD's.
Right now I'd expect a larger number of HDD's to be failing simply because they would tend to be (But not neccesarily be) older than your average SSD making them have a greater chance to fail within the next year, two years.... Whatever timeframe you'd like to use.

To me it's the same comparison of saying you'd expect a model vehicle made in the 1990's to fail before one made in the 2000's AT THIS TIME, not that one is really any better than the other in terms of expected failure rates or for how long they'll go before a failure occurs.

The 90's models have been around for a longer time vs the 2000's and chances are with more wear and tear (Miles) too, that's why I say it related to this.