Trying for glory a second time, Seagate is very soon to announce a hybrid solid state and spinning disk drive, the Momentus XT, with no operating system dependencies. It marries a 2.5-inch, 7200 rpm 250, 320 or 500GB, 3Gbit/s SATA hard disk drive (HDD) with a 4GB solid state drive (SSD), presumably a Seagate Pulsar drive, with …
Why only 2.5"? The price/performance ratio seems to be really good. I wouldn't mind this on a desktop workstation.
Of course there's not problem in putting one of these drives in a desktop. However, if the issues is capacity, on a desktop then I think a better value approach, (and better performing for most purposes) as there are usually spare bays, is to put the system and commonly other commonly used space onto a suitable SSD (the Kingston SSD Now 64GB is a good choice) and put things like bulky media files onto a 3.5" HDD.
A 64GB SSD and a 1TB drive can be had for about £190. If you have Windows 7 you can use symbolic links to mount subdirectories from the 1TB drive into a MyDocs on the SSD so it is seamless (prior to Windows 7 you have to make do with mounting whole partitions into sub-directories, but that's still workable). It may not be quite as clever as this solution in that it doesn't work out access patterns, but it gives over 1TB of storage and 64GB of that is SSD.
For most users it works like a dream, and if you chooses your SSD wisely then even random writes perform a bit better than an HDD.
Linux users can, of course, do this stuff with ease.
I'd buy that.
Actually, I'd buy several...
Ehmm, I think the colours are kinda reversed.
If the article numbers should make any sense the orange WD 250 should be the HGST, right? And the other colours are probably off as well.
Other than that the drive looks nice, though I would still opt for a clean SSD drive for system and anything else that needs to be FAST and a standard cheap SATA drive for storage.
Sorry, my bad. Read the names wrong at first, d'oh.
A quarter of SSD prices
but *almost* SSD performance?
I'll have two in RAID 0 then.
That write strategy
SSDs are often quite poor on writing, so I can imagine why they've followed a "write to the disc first" policy, but this does mean that the device probably can't spin down (as suggested by one person in the article) unless there are absolutely no writes at all. That's unlikely.
Also, write performance is not *that* critical, since in most cases the OS's cache will hide the latency from applications. (You'd only notice if you were streaming more data than the cache could hide.) It's not like reading, where the applications are actually stalled until the data turns up.
So have they just thrown away a Unique Selling Point in pursuit of a slight increase in a benchmark that doesn't matter much?
Sounds pretty cool
If this works with Windows (I assume that's where the testing took place), then I can't wait to try it out with Ubuntu! :-D
I certainly would not buy it with Seagate written on the label. In 15 years of IT I have seen a hugely disproportionate high failure rate of Seagate HDD's specifically. Steer clear if you ask me.
Hang on until Samsung make one.
Now I'm glad...
when I bought my Sony i7 notebook that I took the default, cheap 5400 RPM drive, rather than spending £90 to upgrade to the 7200 model. NOW I have £90 towards one of THESE puppies...I want it, I want it!!!
>>The market aim is to sell the drive into gamers' workstations and high-end notebooks
Great for gamers because it's only just a bit worse than everything else? (if the crysis graph is anything to go by).
I'd be interested about boot from hibernate times but I'm not sure that saving 15 seconds on boot is worth the extra.
Why all the excitement about the data storage algorithm? It sounds like a boring old write through cache to me. I once owned a caching controller that took a power feed off the PSU and did what this drive does.
It's nice that it doesn't need some form of power backup and is integrated into the HDA but it hardly seems revolutionary.
The integrated SSD doesn't optimise writes - it optimises reads.
Frequently read sectors are transparently mirrored on the SSD. Result: frequent 'random read' patterns (such as system/application starts) are cached on the SSD.
Hasn't this been around for a while?
"HDDBOOST" looks to offer something similar by combining the HD/SSD yourself. I was tempted by one of these a few months back just out of curiosity!
One question, why would they worry about stuffing SLC into this item when their cache is meant as a read-cache? Sure, it will learn over time to stuff different info in the cache, but it would hardly change as much as a pure SSD would to worry about write-endurance. And since they're writing to the HDD first, they could hardly be concerned about write-latency (which is the other primary reason for SLC). Lastly, WTF only 4GB??? Win7 is twice that installed, not to mention if you stuff Office and a few choice games on there. That cache would fill up quite quickly. I would buy one if it had a 3.5" version with an 8-16GB cache.
Get that functionality much cheaper
The open source zfs handles addition of SSD as cache automatically for you. You just build up a zfs raid with lots of discs. Then you add a couple of SSD writing one command and zfs handles them automatically. No need to buy special hardware with two drives and complex logic which may fail. zfs is safe and extensively tested, I doubt this Seagate disc's is as flexible as zfs solutions. You can add any SSD as cache, very fast, or very large, in the future without problems to zfs. This Seagate SSD is only 4GB, it is too small.
There are zfs setups (raid + SSD) which gives 300.000 of IOPS and several GB/sek bandwidth, read link below. And you can just replace discs individually if they break. You can not replace one disc in this solution. This Seagate solution is more expensive, inferior and slower than free open source zfs solutions. What's the point?
please read more carefuly
This is about 2,5" laptop drives. So your post is relevant only if you manage to cram a RAID setup into a lappie and then install Win7 on a ZFS partition.
Look on the bright side ....
Probably won't be long before Western Digital come up with something similar.
Seagate - just say no!
Real World Performance?
A synthetic showing "close" performance to an SSD? Did the testing involve allowing the drive to optimise for a particular test? What would performance be like after a few months of use and a user who randomly uses quite a few applications ... what is the performance in this situation?
Hey, I really hope they work as well as they say they do, but just not sure this proves they do (4gb is pretty small). At the moment the suggestion to have an SSD program drive and a HD data drive sounds the ticket, but that is still a very expensive option for the desktop.
@Kebabbert - ZFS is relatively dead isn't it, particularly on the desktop? Mac - nope. Windows - nope. Linux - nope.
I wonder how this device would cope with RAID??? I mean what if the files are on the platter on one disk and Flash memory on the other?!
Gorbachov, Anthony 13
Gorbachov, to be picky, technically you can install Win7 on top ZFS, if you use SAN solutions. Use a diskless PC, and boot from the network where your ZFS server is. On your ZFS server, you give an ZFS partition to the diskless PC client and install Win7 on it. Then boot. With this solution you have the safety from ZFS (NTFS can corrupt your data) and can also use ZFS functionality, such as snapshots, etc.
But regarding this 2.5" drive, I dont see the point of using this hybrid instead of a normal SSD?
Anthony 13, I dont know about ZFS user ratio on desktops, but maybe you know.
But I know that ZFS is very live on Enterprise servers, because it protects your data, whereas XFS, ReiserFS, ext3, etc - all can corrupt your data, without even noticing it. This data corruption also occurs on desktops of course. Therefore I think ZFS is the best solution even for desktops, because your data is safe. At last.
Netapp PAM replacement
Use an enterprise drive combined with this tech and everyone gets access to a Ntap PAM equivalent ?