Slashdot Mirror
Seagate Bulks Up With New 8 Terabyte 'Archive' Hard Drive
MojoKid writes Seagate's just-announced a new 'Archive' HDD series, one that offers capacities of 5TB, 6TB, and 8TB. That's right, 8 Terabytes of storage on a single drive and for only $260 at that. Back in 2007, Seagate was one of the first to release a hard drive based on perpendicular magnetic recording, a technology that was required to help us break past the roadblock of achieving more than 250GB per platter. Since then, PMR has evolved to allow the release of drives as large as 10TB, but to go beyond that, something new was needed. That "something new" is shingled magnetic recording. As its name suggests, SMR aligns drive tracks in a singled pattern, much like shingles on a roof. With this design, Seagate is able to cram much more storage into the same physical area. It should be noted that Seagate isn't the first out the door with an 8TB model, however, as HGST released one earlier this year. In lieu of a design like SMR, HGST decided to go the helium route, allowing it to pack more platters into a drive.
you are better off with generation-1 than generation-current.
never trust the very leading edge. and, we're talking seagate, here; their enterprise drives are ok but I wouldn't touch them, these days, for consumer drives. no way!
no way I'm trusting helium, either; since it escapes and makes the drive useless a few years down the line.
--
"It is now safe to switch off your computer."
you are better off with generation-1 than generation-current.
never trust the very leading edge. and, we're talking seagate, here; their enterprise drives are ok but I wouldn't touch them, these days, for consumer drives. no way!
no way I'm trusting helium, either; since it escapes and makes the drive useless a few years down the line.
But you'll be able to tell when that happens when your voice gets really squeaky.
and then let's hear about how it's all anecdotal evidence.
Then someone will bring out the backblaze survey.
Then someone will say "They've never had a problem with Seagate, but WD sucks."
Then someone will lament how IBM no longer makes drives. Then the deskstar stories will start.
In other words, the same responses every time a hard drive story is posted.
> So...it's magic?
Could be. Or, it could be sufficiently advanced technology-- it's hard to tell.
"Reduce friction" is pretty close, actually.
The platters spinning around causes a lot of air to move around, as well. If that air is helium, the effects of the turbulence are less forceful, so moving parts don't need as much buffer space between them.
The individual platters don't change density, but since they can be packed closer together without aerodynamic damage, there can be more platters in a single unit.
You do not have a moral or legal right to do absolutely anything you want.
No, straight Helium is a lower density medium than normal air. This means less atmospheric friction and less driver motor friction while spinning a platter.
Since the individual platter assemblies run cooler, they can pack them closer together (and put more in a given drive casing).
Also, because they have to hermetically seal a platter assembly into the helium atmosphere, with some modifications, such drives can be used in full-immersion cooling, where normal air-cooled drives need to breathe.
Chas - The one, the only.
THANK GOD!!!
easily fixed with autotune...
--
"It is now safe to switch off your computer."
There are only 3 hard drive makers left. Hitachi is not one of them.
their enterprise drives are ok but I wouldn't touch them, these days, for consumer drives. no way!
There is no difference in reliability between "enterprise" and "consumer" drives. Those are purely marketing terms. The sole advantage of enterprise drives is a longer warranty. If you are bad at math, you might think that is a good deal.
Why does this drive have the archive moniker? Is it any more reliable than a non-archive drive? The name suggests I can put data on it and shelve it for 20 years and come back with all the data still there. Is there any indication that might be the case? Somehow I doubt it.
-- Thou hast strayed far from the path of the Avatar.
fucking helium, how does it work
Not entirely correct:
http://www.theregister.co.uk/2...
They are owned by WD, but they are still a manufacturer in their own right. For the moment.
"We know what happens to people who stay in the middle of the road. They get run over." - Aneurin Bevan
I got a Seagate 3TB in a USB enclosure a year or two back.
Worked great for a year to a year and a half, then I started getting it randomly hanging. At first I assumed it was the usb interface going back, but upon removing the drive and directly plugging it into the system the same symptoms remained. Since it had been in an enclosure it hadn't supported SMART access to the drive. The SMART readings with the bare drive didn't show anything obvious, but actually reading from the drive would give read errors, and too many read/write errors over a certain period would cause the drive to hang, sometimes hanging the entire bus.
Long story short, it turned out I wasn't the only one having this problem, it happened pretty commonly across that entire serial line of drives, and there was neither firmware fix, nor warranty support for them (The enclosures only gave a 1 year warranty despite the drives having a 3 year warranty tag printed on them. The only thing I can figure is they figured out the entire batch was bad, about how long they'd last, and shoved them in a bunch of USB cases where they didn't expect anybody to find out.)
Having dealt with that drive, and reviews of them online, I'm going to be aversive to those, hitachi 3tb, and possibly WD 3tb for the forseeable future. Knock on wood, I haven't had ANY problems with 2 terabyte drives so far and given that another stepping of drives is coming out, we might see the later versions of the current-gen drives becoming mature enough to rely on for more than a year, which going off reviews doesn't seem statistically safe yet for this generation.
SMR drives are fine for I/O scenarios that don't overwrite data very often but they suffer from significant performance penalties for overwrites due to the read-modify-write/write-relocate operations required to modify a set of sectors within a shingled-encoded track. There are tricks to lessen the impact such as virtual sector remapping and background remapping but those can't avoid performance penalties in many scenarios.
Yes, I also have very old Seagate drives with capacities from about 40 to 300 or so gigabytes that work fine. I also have a 5 gallon pail full of dead 1 terabyte drives that are 2-4 years old. I do IT consulting (mostly for small business) and the failure rate on the 1 terabyte and up drives has been hideous. I have been hammering on all my customers to do full drive image backups regularly - and to replace the backup devices as soon as they are over two years old. I'm generally not a hard sell guy, but I am pushing this, because I don't want them to be able to say they weren't warned when I have to charge them $thousands to get going again after a drive fails.
Crow, listen to this guy. Assuming these things have 100MBytes/sec write speed, a simple RAID-1 will take over 22 hours to rebuild.
If you want 8TB of usable space, get 4x4TB and RAIDz2 (i.e. RAID6) them. Even if it's disposable data, the data must be of sufficient use to justify a FreeNAS build over a simple external. It's worth your time to do it right.
I work for a very large storage array manufacturer. Warranty length is *not* the only difference...
Agreed. The price is also different.
For reliability, I prefer actual data over your anecdotal opinion: Consumer drives shown to be more reliable than enterprise drives.
You mean you got hit by the 7200.11 bug and didn't do any research into it to discover that it's a firmware issue with a simple fix?
Unfortunately there is a common trend in the commercial world where a once-quality brand decides to cash in on it's reputation and sell low-quality crap and "We're a quality brand" prices. No doubt it boosts profit margins dramatically, for a while, but means the world loses another quality brand, and a lot of customers get screwed over. And sometimes it's a graduated process where the high end enterprise/boutique products continue to maintain their quality to prop up the brand, while the quality of the normal products falls off a cliff.
I haven't been following hard drives closely enough to be able to comment on Seagate's case, but I've seen it happen to far to many once-great brands to be even remotely surprised.
--- Most topics have many sides worth arguing, allow me to take one opposite you.
Don't use RAID5 with drives over 1TB.
a) a RAID5 rebuild takes many hours, b/c it involves reading the entire disc.
b) drives from the same production batch tend to cluster failures.
c) I recall reading that the uncorrectable read error rate tends towards the 2TB mark.
That is, chances are very good that a single drive failure will become a 2-drive failure during a rebuild.
RAID6 or nothing.
You mean you got hit by the 7200.11 bug and didn't do any research into it to discover that it's a firmware issue with a simple fix?
Its simple to upgrade the firmware when you can still access the drive otherwise you have to jigger up a TTL level serial interface and send AT commands to unbrick the thing...lots of "fun".
Write Once Read Mostly
Shingled media is almost useless for random access, since rewriting a logical block means relocating its entire "shingle" strip somewhere else., then, at some other time, garbage-collecting the entire region and relocating the still-in-use blocks. You definitely want to run these "noatime", to prevent thrashing directory blocks, and they should probably have a new filesystem designed for them.
Some have tried tinkering with flash filesystems due to the "copy/invalidate/garbage collect" and the LBAs are gathered in some larger storage block in no particular order, and that storage block needs to be managed. Don't know if Seagate will tell us what the size of a erase block (a set of overlapping, concentric "shingles", which have to be collected as a group) really is, or if they'll even be a consistent size.
If you're streaming from them, you may hit "garbage collect" long access times, and I don't know what proprietary commands and settings may be available, if any, to tell the drive "now is a good time to do housekeeping".
As "archive media", shingled drives probably work OK, since that is a WROM application, but, personally, I would NOT use them on any existing file system.
No. Go look at an upper mid-sized enterprise, and ask what kind of hardware they have running their Microsoft SQL Servers, their Exchange server, or their Oracle cluster.
What Google, Facebook, and Yahoo are doing is not relevant at the enterprise level. These are super-colossal cloud-scale companies, that are 3 orders of magnitude larger than Enterprise computing, not ordinary enterprises.
Enterprise hard drives are designed for Enterprise use, not Google or Facebook's cloud or HPC clusters.
These massive companies also have their own custom hardware built at their disposal. They are not using RAID arrays like most enterprises are using, and they essentially have massive farms of workstations instead of servers running their computational workloads.
At sufficient scale, you can achieve reliability from consumer disk drives for in-house applications, by designing your application around your components, BUT the major requirement is that you are in control of the application stack, so you can actually use the disk drives like you want --- and not have to stick them in a tightly-coupled RAID array.
The consumer disk drives are not sufficiently unusable that you can't work around the limitations by having thousands of them in a cluster, with terabytes of cache spread over 5000 computers, and some smart application logic doing what ordinary RAID subsystems cannot.
You mean you got hit by the 7200.11 bug and didn't do any research into it to discover that it's a firmware issue with a simple fix?
So you bought the Seagate company line about that? Either you never owned one of those drives or you were one of the lucky few that was eventually helped by the firmware fix. Although why you would wait around for many months for the 'simple fix' when you could get a refurb replacement immediately I don't know.
This is why a good PR firm is worth its weight in gold. It's okay to have a catastrophic production failure as long as you can retroactively convince the ones who didn't get burned that it was all just a big misunderstanding and was easily fixable with a simple firmware update. If only Hitachi had done so well with their infamous Deathstar drives.
So you believed their propaganda. Go back to the Seagate forums from that time and I think you will see that the so called "firmware fix" only fixed a small percentage of the problems with those drives. There was another fix that helped some people (more than with the firmware update) that involved removing the pc board of the drive and hacking the hardware yourself. I believe a soldering iron may have been required in addition to a particular sort of cable. I can't remember exactly but it was not a fix that most people would be able to apply and often it didn't work anyway. I had a 1.5 TB 7200.11 that I had been keeping for ages to eventually buy the cable and apply the fix but by the time I got around to maybe doing it 1.5 TB was a very small drive and I didn't care so much about the lost data anymore.
I had 6 7200.11s. Both 1 TB and 1.5 TB. Most failed in less than 6 months and then their replacements failed too. None of them work today. Not a single one. And your firmware fix could not be applied to any of my drives because it was not a firmware problem. At least with my drives. Yes a small percentage of 7200.11s did have firmware problems, but mostly it was a hardware unreliability problem. The click of death as well as drives that just refused to stay online for long. They'd just drop out. And all kinds of 'delayed write' errors etc. Those were not caused by poorly written firmware. They were 100% authentic hardware problems and Seagate shipped out countless new drives to replace the things on warranty which would seem like a rather expensive thing to do if all they had to do was update the firmware. But maybe you will say even seagate "didn't do any research" and was unaware of the "simple fix" you speak of.
Despite your convenient assumption about lots of 7200.11 owners being unaware of the too little and far too late 'fix' of a firmware update that didn't even work for most owners, I suspect that most found out about it when their drives started failing. A simple google search for '7200.11' and 'clicking noise' would eventually have gotten hits for the so called 'fix'. Of course it took Seagate forever and a day to even come up with that. I don't think they have ever even admitted that there was any sort of problem with the drives and by the time they came up with your so called "simple fix" most owners had already been burned pretty badly by their decision to go with Seagate. Before my 7200.11 I had been a big fan of Seagate. Nearly all my drives were Seagates. Now I don't care what name is on the drive. They are all incredibly unreliable. I have better luck with their refurb replacements usually.
Quite an experience to live in fear, isn't it? That's what it is to be a slave.