In The Lab With Seagate's Momentus XT 750 GB Hybrid HDD

[neiroatopelcc]

[citation][nom]Article[/nom]Once you've used an SSD, it's hard to go back to a system armed with a hard drive.[/citation]

So bloody true. Company laptop's equipped with a C300 SSD drive. Doesn't feel very quick. But once I get back home and fire up my 'highend' system still running on a hitachi 7200rpm drive it feels very very very slow in comparison despite not really being slow.

Maybe I should get one of those seagate drives as a system drive and free up a 2TB drive. I need to buy a new drive for my server anyway - might as well just recycle one of the old drives.

Not too sure about seagate's 0.5% AFR though. I've got drives from WD, Samsung (pre seagate) and hitachi in my systems at home to avoid a disaster like the 1TB seagate firmware. I wonder if adding seagate to my system would cause my system to one day just not boot up...

ps. I would guesstimate, based on my work enviroment, that the seagate fail rate is more like 1-1.5% per year. Almost all harddrives we're seeing failing in HP desktops are seagate. Granted they're also teh most common make, but I don't recall the last time a hitachi drive failed, and I think I've only replaced one WD drive in the last couple years.

 

[richard hart]

There are some interesting statistics on RMA’s on a French web site that are based on statistics from a major retailer. Apologies, but it is in French. It shows a wide variation and it does not include products that were sent directly to the manufacturer for RMA.

Key highlights:

SSD -
- Crucial 0.82
- Intel 1.73%
- Corsair 2.93%
- OCZ 7.03%

The highest return rate was for the OCZ Vertex 2 Series 240GB drive, which came out at 15.58%
http://www.hardware.fr/articles/862-7/ssd.html

HDD –

- Samsung 1,23%
- Western 1,63%
- Seagate 1,89%
- Hitachi 3,95%

The highest return rate was for the Seagate Barracuda XT drive, which came out at 7.53%

http://www.hardware.fr/articles/862-6/disques-durs.html

There are no figures for Hybrid drives in this report. When a SSD fails the possibility of data recovery is extremely low, however this is not usually catastrophic as they are typically used for OS/ apps that can be easily restored. The possibility of recovering data from a HDD/ HHDD is reasonable and this is an important factor, as they are typically used to store important data.

Cheers,
Richard Hart
TomsHardware.com

 

[richard hart]

Quite a few people have asked me about how the Momentus XT works with games. Games typically consist of a workload that is more dependent on sequential file xfers, rather than small random file xfers that are more likely to be held flash.

The only game I play at the moment is MW3 MP, which is slower at loading than a SSD. The result is that I have less time to wait for other players to become ready before a game starts. Once the game has starts I can’t tell any difference between a SSD and the Momentus XT. Games vary however so YMMV.

Cheers,
Richard Hart
TomsHardware.com

 

[neiroatopelcc]

Just placed an order for one of these drives. Hope I get the same positive experience as you despite my (subjective) dislike for Seagate drives.

 

[wsandman]

So would this be a suitable replacement for a desktop system instead of a laptop?

 

[andy_newton]

I don't care how cheap, how fast, how cool.

Seagates are designed to die within less than 3 years.

If yours survived 3 years then what you have is a malfunctioning unit.

 

[blazorthon]

[citation][nom]americanbrian[/nom]I made 2 statements and I stand by them. No.1 It is LIKELY that these drives will be less reliable...Due to the laws of probability, I believe the spider graph is misleading. Unless they make both the HD and SSD components 2x as reliable as normal. I don't think this is realistic, they will be made the same as current tech allows.No.2 Probabilty does work like that... (when applied to large sample sizes with which were being discussed)I do not see what is overbearing about either of these statements. That is all.[/citation]

Your statements are overbearing because they function under the assumption that these drives must be less reliable, on average, purely because they have both flash and hard disk storage. First of all, the flash isn't an added chance of failure because if the flash storage fails, then these drives still work... They simply work as regular hard drives instead of as high performance hybrid drives. So, their reliability must be at least as good as a similarly built hard drive. That there is flash does not increase the chance of failure because the flash is not necessary for the drives to function.

So, all that these drives need in order to be more reliable on average than most conventional hard drives is to be built above-average for a hard drive. You're trying to apply math in ways that it doesn't work in because you don't know the context.

 

[kyuuketsuki]

[citation][nom]blazorthon[/nom][...]if the flash storage fails, then these drives still work... They simply work as regular hard drives instead of as high performance hybrid drives. So, their reliability must be at least as good as a similarly built hard drive. That there is flash does not increase the chance of failure because the flash is not necessary for the drives to function.[/citation]
Oh good, someone who actually knows what they're talking about came into the discussion.

Also, the failure rate for the flash portion of the Momentus XT is not going to be comparable to general SSD failure rates (which are highly overestimated anyway). The flash portion of the drive is SLC, not MLC, first of all. Second, most SSD failures seem to be firmware related and not due to the actual flash failing.

 

[bluekoala]

Wow.
willard, you have a lot of good skills.
You should join our lucrative buisness by applying here:
http://www.collegehumor.com/picture/6705045/do-you-recognize-this-couch

You'll get to lay on a couch rather than an office chair.

 

[americanbrian]

Your statements are overbearing because they function under the assumption that these drives must be less reliable, on average, purely because they have both flash and hard disk storage. First of all, the flash isn't an added chance of failure because if the flash storage fails, then these drives still work... They simply work as regular hard drives instead of as high performance hybrid drives. So, their reliability must be at least as good as a similarly built hard drive. That there is flash does not increase the chance of failure because the flash is not necessary for the drives to function.

So, all that these drives need in order to be more reliable on average than most conventional hard drives is to be built above-average for a hard drive. You're trying to apply math in ways that it doesn't work in because you don't know the context.

I would argue, as I mentioned previously, completely aware of the context that flash failure constitutes drive failure as it doesn't perform as advertised any longer.

I said that previously. Just because there is no data loss, does not, a working hybrid drive make.

I take it the context is reliabiliity in terms of data integrity, as opposed to functioning as it needs to for your application.

I am specifically speaking of the people who really leverage the caching and probably buy multiple units. Like a previous commentor who found that an array of these was exactly suited to his purpose. If the falsh fails it would no longer suit his purpose. As in, he will have to buy another to replace it to maintain functionality of his array.. THIS is what will happen more often with these drives.

I think you are being really offensive. I will no longer be responding to your posts.

 

[blazorthon]

I would argue, as I mentioned previously, completely aware of the context that flash failure constitutes drive failure as it doesn't perform as advertised any longer.

I said that previously. Just because there is no data loss, does not, a working hybrid drive make.

I take it the context is reliabiliity in terms of data integrity, as opposed to functioning as it needs to for your application.

I am specifically speaking of the people who really leverage the caching and probably buy multiple units. Like a previous commentor who found that an array of these was exactly suited to his purpose. If the falsh fails it would no longer suit his purpose. As in, he will have to buy another to replace it to maintain functionality of his array.. THIS is what will happen more often with these drives.

I think you are being really offensive. I will no longer be responding to your posts.

Even in that case, as has been said, these drives have much more reliable SLC flash instead of the cheaper MLC cache, so their flash should be far more reliable than that of most SSDs. The chance of the flash failing should be much lower than the chance of the flash failing in a standard SSD which is already low. These drives also don't have a poorly implemented SandForce controller, so the chance of the cache failing at all is slim. So, like I said before, their chances of failing are pretty much as good as the chance of the hard drive portion of the drive failing. Their reliability is more dependent on how well built the hard drive is. You can argue whatever you want to argue, but this is a moot point and can't be solved any further without proper evidence. You're trying to argue that these drives are less reliable than any other without any proof of it and whether any of us likes that or not, I think that this argument is not only flawed, but also unreasonable.

Having read through my previous comment, I disagree with what you said about it being offensive and I don't see how you found it to be so. If you don't want to respond to me, then fine; that is your choice. However, I was not and did not intend to seem offensive.

 

[bluekoala]

I find it amazing how much some people can get so emotional over hypothetical drives. The comments of this article is like a nerd soap opera. It's strangely stimulating. Please don't stop, I'm wondering what's going to happen next.

 

[reoowens]

I don't understand why such an old regular hard drive was used, was this to drive the speed difference point home? Why not use something a little more modern, like a Velociraptor? Were you worried that the results would be too close to make a good case for the hybrid?

 

[Shin-san]

The #1 thing keeping me from getting a hybrid HDD: the things come with DRM that locks the hardware. I don't want to have to troubleshoot some lousy DRM scheme. Otherwise, I would love to get one for my laptop or for secondary things on my desktop like games.

For me, most of the time I care about is loading programs, dlls, and etc. Large files like music and videos usually don't matter too much because they don't need the I/O bandwidth. The only time I'd use large files and the I/O at the same time is video capping, and that's not used that often on my laptop. Also, large files tend to perform well off magnetic (not as well as SSD but I'm more worried about 4k performance).

Edit: seems like people aren't reporting DRM problems for Seagate hybrid drives

 

[dthx]

[citation][nom]reoowens[/nom]I don't understand why such an old regular hard drive was used, was this to drive the speed difference point home? Why not use something a little more modern, like a Velociraptor? Were you worried that the results would be too close to make a good case for the hybrid?[/citation]
Quite logical in fact. This drive is clearly designed with laptops or compact machines that can physically only accept one 2.5inch drives in mind. The goal is to offer a balanced solution: an affordable drive with a substantial but not extreme performance boost without sacrifying capacity.
If money wasn't a problem, a 960Mb SSD would work out much better for you and in a tower case, a small SSD + a large HDD is a better fit, but you don't need a test to know that. Regarding the raptor, I know the disc is 2.5 inch but the drive casing/heat dissipator is 3.5 inches big so it doesn't fit in a laptop (and it would kills the laptop's batteries in no time, overheat and make too much noise).
By comparing this drive with an old regular drive, the author gives an excellent idea of what potential buyers can expect from such kind of upgrades.

 

[axiome]

If anyone is interested in the controller tech used here, I found this blurb off legitreviews:

You might be curious if Seagate is doing any wear-leveling on this Micron SLC NAND Flash chip and we were told that they are not. In fact here is what their engineer had to say when we asked them about over-provisioning.

"Unlike SSDs the Solid State Hybrid while using all of the 4GB, it does not have to. Since we do not use the Flash as primary storage, but instead are using it as a temporary cache, then as blocks wear out we just use fewer blocks for the caching. We are able to achieve our performance for most users without using all of the 4GB, but we will evenly use all of the Flash as we move data in and out. With all of that said, since wear out of the Flash occurs with writing to the Flash the Solid State Hybrid does not wear as much as a SSD because most of the write data goes to disc first and we only move data that is read frequently to the Flash. This reduces the number of writes to the Flash. And one final note, an SSD will have write amplification due to background 'garbage collection' of partially used blocks. This is necessary on the SSD to retain the write performance by maintaining a certain amount of erased blocks. Because the Solid State Hybrid also has the resources of a disc our Flash file system has very little write amplification." Seagate Engineering

 

[superp]

can anyone confirm that the "spin-down" problem has been solved with this new release or not?

 

[mynith]

But is it cheaper than a hard drive, 8 GB of extra RAM, and some clever software to use all unused RAM to cache the hard drive? I think not. And at the end of the day, you still have the extra RAM on those rare occasions that you actually need all of it.