Hitachi Ultrastar A7K2000 2TB HDD Back in August 2009, Hitachi announced that it was “shipping the industry’s first 2TB 7200rpm desktop hard disk drive” in the shape of the Deskstar 7K2000. Hitachi Ultrastar A7K2000 Hitachi's Ultrastar A7K2000: better duration specs than the Deskstar That statement may have been accurate, depending on your definition of ‘ …
People keep stating that the public should use SSD as their boot/system drives. SSD still has fundamental failings in the number of writes it can survive, and you will see failing SSD drives in large numbers.
People should offset tmp and other system written files where possible to HDD drives if they use an SSD.
@By Anonymous Coward Posted Tuesday 24th November 2009
Read the article, not just the headlines. It's above the pretty picture.
"The nominal capacity of 2TB yields a true 1863GB once the drive has been formatted."
Next have to agree about 5 year warranties, nice to to have, but a waste of time, after 4 or 5 years the shipping often costs more than the drive is worth.
"The question is how much emphasis you put on the performance of your hard drive. If you’re looking for speed – and who isn’t? – you really should have an SSD for your boot drive. This leaves the 2TB drive as a storage drive where performance is less important than it might otherwise be..."
Umm, not so fast there. Speed might not be as important for storage drives in general, but for some applications it's every bit as (if not moreso) important as the speed of your boot drive. I suspect pretty much everyone who works with large files has spent many idle moments wishing the load/save progress indicators would move just that little bit faster.
And if you're building a small form factor PC where you don't have the luxury of being able to fit a small and speedy boot drive alongside a larger slower storage drive, then getting as much performance as possible out of your single multi-role drive is quite handy.
You completely fail to understand wear levelling and over provisioning. SSDs dynamically redistribute sectors between different physical locations each time they are written. Because the time to fetch a sector is independent of it's location it doesn't matter if a file is scattered across the physical flash memory. As there are more physical sectors than are reported and ECC logic is applied to each sector the firmware can identify problem sectors long before they cause an issue and map them out while still providing the same capacity.
The result of these two factors is that you could hammer a modern SSD with write data for 5 years continuously and it would still be fine. In practice they are MORE reliable than HDDs as they are less susceptible to heat, vibration and mechanical wear.
Big HDDs like the Hitachi will be around for a while yet for bulk storage, but SSD is rapidly going to become the preferred technology for system disks and fast data.
If you refer to our review of the 40GB Kingston SSD you'll see that performance was lower than the 128GB SSDNowV that we previously reviewed. We speculated that this is a result of the 40GB using five memory chips which doesn't take full advantage of the ten channel memory controller.
So yes, your 64GB drive is very likely faster than the 40GB.
"SSD still has fundamental failings in the number of writes it can survive, and you will see failing SSD drives in large numbers"
http://www.storagesearch.com/ssdmyths-endurance.html
"People should offset tmp and other system written files where possible to HDD drives if they use an SSD."
Agreed
@AC 13:34
"So how much actual disk space do you get from it ? 1 TB ?"
Well, somewhere close to the 1863GB it actually says in the article.
Also..
"This Ultrastar drive is horribly expensive and delivers little in return. However, its low-cost Deskstar sibling looks incredibly tempting."
And yet it gets a 60% rating? Does it need to create a black hole, abduct children and be a terrorist in order to get <50% rating?
Everything has a MTBF by definition. Either a) they can't be @rsed to measure it or b) they measured it and they're hiding the results because it's "2" or something.
I really really want one, I do, but looks like I'll just have to extend my array with 1TB drives for the moment...
@DS1 - "People should offset tmp and other system written files where possible to HDD drives if they use an SSD"
Sorry I just couldn't let this one pass without comment. Don't you think that would rather defeat the point of using an SSD in the first place?
To be fair, the deathstar era was a very long time ago now, and since then the drives have been fairly good.
I'm currently running 13 HDs in my various PCs at home, 10 of which are HGST, and they're the most reliable of the lot. I've even got a deathstar in my xbox (which did fail, but has since had the firmware upgrage) and even that's still going strong.
6 of these are in my home server, which is on 24/7. 4 are in a RAID array, and SMART says they've currently been powered up for something like 1.8 years. (16153 hours), can't praise them highly enough.
I have an SSD as my sysem disk. Admittedly a rather expensive 256GB Crucial one, but I'm not very worried about write endurance because, if you do the maths and have a typical PC usage pattern and the wear leveling is even vaguely successful, it's going to be several decades before it's a problem (and the drive also has a 5 year guarantee). In any case, HDDs wear out and fail and I have suffered that. Most likely my SSD will outlast an HDD (people mistake MTBF figures for average lifetime of disks - very, very different things).
What I can also say is that the SSD has totally transformed the performance and usability of the PC.
Of course you do need an SSD with proper wear-leveling (don't use a USB stick), and I wouldn't recommend a consumer type drive for a volatile enterprise database. I'd also recommend not using the whole disk - whilst SSDs don't suffer big performance problems through file fragmentation, they do benefit from having unused space due to the way that writes have to be performed.
"people mistake MTBF figures for average lifetime of disks - very, very different things"
Given that
MTBF = mean time before failure.
Failure = end of life
mean = sum of samples / number of samples = what most recognise as the average (rather than median or mode)
Unless you are asserting that failure does not eol the device, I fail to see what mistake people are making. Or are you suggesting that mtbf values are calculated/extrapolated rather than measured and therefore wrong...
"The nominal capacity of 2TB yields a true 1863GB once the drive has been formatted."
WRONG!
Once formatted the drive still yields 2000GB (gigabytes). In fact it was 1863GiB (gibibytes) even before it was even formatted!
2000 GB = 1862.637 GiB
2000x10^9 bytes = 1862.637x2^30 bytes
I know I know, who cares? And I agree it is pedantic of me to correct this but software AND hardware manufacturers need to start to adopt this convention to minimize consumer confusion. Remember that Mars probe that NASA lost? UNITS!!!
See http://www.storagesearch.com/ssdmyths-endurance.html for myths about SSD "Write Endurance". You're talking decades before current, 64+GB SSDs reach their write limit. Compare that with the number of mechanical HDD failures we've all experienced over even the last 5 years.
I'm writing this on a SSD equipped MacBook Air so I hope the author got his sums right :)
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2024
