Western Digital Starts Shipments of HGST Ultrastar He12 12 TB HDDsby Anton Shilov on April 28, 2017 10:00 AM EST
- Posted in
- Western Digital
- Helium HDD
Western Digital on Wednesday announced that it had begun to ship its HGST Ultrastar He12 hard drives with 12 TB of capacity. The HDDs are the first drives to employ eight platters, so the fact that Western Digital is now shipping them is important not only for its datacenter customers who need massive storage capacities, but also because the drive represents a significant step forward from a technology point of view.
The HGST Ultrastar He12 is based on Western Digital’s fourth-generation HelioSeal technology, which uses eight perpendicular magnetic recording platters with 1.5 TB capacity each. To add the eighth platter, Western Digital had to redesign internal components of its HDDs (including arms and heads) significantly. In addition, the company increased areal density of the platters, which improved the sequential read/write performance of the new hard drives. In particular, Western Digital claims that the HGST Ultrastar He12 has a sustained transfer rate of 255 MB/s, an average latency of 4.16 ms, as well as an average seek time of around 8 ms.
Just like other enterprise-class nearline drives, the Ultrastar He12 features a 7200 RPM spindle speed as well as all the technologies currently found in such hard drives from HGST, including a special micro-actuator that improves the accuracy of head positioning in multi-drive environments (which naturally improves performance, integrity, and reliability), rebuild assist mode to speed up RAID recovery time, and others. Finally, the new drives have SED options as well as Instant Secure Erase feature so to either quickly redeploy or retire a HDD.
|HGST Ultrastar He12 General Specifications
(12TB model Specifications only)
|Interface||SATA 6 Gbps||SAS 12 Gbps|
|DRAM Cache||256 MB|
|Format: Sector Sizes||4Kn: 4096
|4Kn: 4096, 4112, 4160, 4224
512e: 512, 520, 528
|Areal Density||864 Gbit/inch2|
|Sustained Transfer Rate||255 MB/s|
|Average Latency||4.16 ms|
|Seek Time (read/write)||8/8.6 ms|
|Power Rating||Idle||5.3 W||6.1 W|
|Operating||7.2 W||9.8 W|
|MTBF||2.5 million hours|
The increase of hard drive capacity by 20% to 12 TB brings substantial benefits to operators of cloud and exascale datacenters because such HDDs enable them to boost their storage capacity by 20% (or more, assuming that they are using drives with lower than 10 TB capacity today) without expanding footprint of their building or significantly increasing their power consumption. For example, a standard server rack can store 2400 TB of data if fully populated with 10 TB HDDs today. Meanwhile, if 12 TB hard drives are installed into the same rack, its total storage capacity increases to 2880 TB at the same power and at the same space.
Since Western Digital’s HGST Ultrastar He12 are aimed at large enterprises, the company does not advertise their recommended prices; the prices customers will pay depends on the number of drives purchased as well as some other factors. Since Western Digital will eventually offer helium-filled HDDs with eight platters to other kinds of customers too, so once this technology trickles down the general public will be able to take advantage of the fourth-generation HelioSeal platform as well.
Later this year Western Digital plans to start shipments of its HGST Ultrastar He12 drives featuring shingled magnetic recording (SMR) platters and 14 TB of capacity. These drives are designed primarily for archival applications, where data is written sequentially in huge chunks (and is hardly ever updated) and where the peculiarities of SMR do not have a major impact on performance. The upcoming SMR HDDs will have to be managed by their hosts and therefore far not all customers of Western Digital are going to be able to take advantage of them.
Source: Western Digital
Post Your CommentPlease log in or sign up to comment.
View All Comments
SaolDan - Friday, April 28, 2017 - linkThats a lot of storage!
ddriver - Friday, April 28, 2017 - linkOn the downside, reliability appears to be gradually slipping ever since the acquisition.
Samus - Saturday, April 29, 2017 - linkThat isn't even true. Why do you think that?
And to imply a slip in reliability would have anything to do with the acquisition, it's timing or whatever, is also bizarre because WD didn't necessarily merge with HGST. Their engineering staff is entirely unchanged. No layoffs or anything were ever announced. As far as the general public knows, and that's us, it's business as usual at Hitachi Storage.
I think it's pretty unanimous since at least 2010 Hitachi has had a reputation of making the most reliable drives and that hasn't changed. Basically the rule of thumb is avoid any Toshiba 3.5" drives and any Seagate Momentus or Barracuda 7200.10-7200.12 drives,
ddriver - Saturday, April 29, 2017 - linkThere are two factors that coincide with the deterioration of reliability - the transition to WD control, which took like 3 years after the acquisition, and the increase of density and capacity It may be either one, or a combination of the two. They don't need to lay off engineers to change the business model, engineers are paid workers, they do as they are told.
HGST drives were impeccable up until about 2 years ago, but the models since then definitely are not up to that standard. Failure rates have skyrocketed to about the same we see for wd or even seagate drives. Seagate was the runt of the litter back then, with wd marginally better, and HGST next to perfect. Since then seagate has actually improved somewhat, wd has slipped somewhat, and HGST has slipped big time.
It makes sense, business wise, to deliberately bring down the production quality, in at least the following aspects:
- HGST drives have a great reputation, and as such, come at a price premium
- making certain compromises with the quality will cut on production cost, allowing WDC to milk the time frame until the declined reliability no longer justifies the price premium, which evidently will take time, as is evident from people like you who are still behind on the fact HGST today is not the HGST it used to be couple of years ago
- it will make wd's own drives look less bad, which they badly need as in the recent years their own reliability has slipped significantly
- since seagate got samsung's hdd business, and wd got hitachi, we are having a defacto duopoly, I am excluding toshiba because their share is negligible, and even in market environments far better than duopolies, we have the implicit price fixing, which in the HDD market also goes into the reliability metrics, the same old "if we all make nearly identical, barely incremental garbage, people will have no choice but to buy"
- it was no coincidence that the lousier HDD makers bought the better HDD makers, samsung HDDs were extremely reliable when the division was acquired, especially while keeping in mind those were really cheap consumer grade drives, I still have one of the first 1TB samsung models running flawlessly for OVER 7 full years, running 24/7 as a scratch disk, 63000 power on hours, about a petabyte of R\W per month, without a single bad sector, without a single smart warning, without making odd noises, which is amazing compared to brand new WD drives I getting, about 20% DOA, another 10% failing the first few days, another 15% failing before the first year.
- letting the reliability slip will not really hurt WDC in the big picture, sure it will likely cause a drop in the amount of HGST drives they are selling, and it will reduce the price they would be able to ask for HGST drives if they want to sell, but all in all, it is not like people can take their business elsewhere, it is either wd or seagate, which is also why I expect for reliability to continue to slip across all vendors, which is not that much of an issue for enterprise because of redundancy, but it is a big deal for Average Joe, who buys a brand new 10 TB expensive HGST drive to put his data on, just to have it result in a tremendous loss
- which brings me to the last reason why reliability is deliberately compromised - the industry really wants people to use "cloud" storage, and they are trying to encourage it every possible way, so the prospect of data loss is a major factor here, "if you don't want to lose your data, put it on the cloud", which is beneficial to the industry for two good reasons - they make way more money renting storage than selling it, they get all of your data rather than having to hack your PC and steal it, so they can data mine it and whatnot, just to figure out how to make more money on you. For the user cloud storage is actually a major inconvenience, especially if you work with a lot of data, and that data is sensitive, because it will cost you significantly more money to store that data, and you will be limited to the speed of your internet connection and the service provider bandwidth, and they all OVERSELL, so they can make even more money on every penny invested.
So yeah, plenty of good reasons for wd to deliberately massacre HGST's impeccable reliability.
jabber - Monday, May 1, 2017 - linkI thin quality dropped from the time of the floods. They all dropped tolerances to get more kit out of the factory and didn't bother to raise them back after things sorted out.
bigboxes - Friday, April 28, 2017 - linkIf only this would bring down the price of the smaller drives. As the makers incrementally eek out the advances it reminds me of Intel in recent years. They milk the profits, but we get little differences in performance/price. It's smart on their part, but not good for consumers. Businesses can't hold off, but I can. However, sooner or later, I will bite the bullet and spend way more than I will have hoped for.
melgross - Friday, April 28, 2017 - linkIt will, at some point. The high R&D costs of these technologies need to be paid off quickly though, so the first devices cost a lot.
A problem for these companies is that we no longer see continued increases in sales. So with higher costs in R&D and manufacture, we get those costs passed along. It now costs about $15 billion for a 14nm chip plant. That's a large part of Intel's sales. Without counting on 6-10% increases in sales per year from new plants, we pay the costs.
haukionkannel - Friday, April 28, 2017 - linkThe HD production is declining so the proce trop is Also getting weaker.
gerz1219 - Friday, April 28, 2017 - linkThe problem is that when we're talking about regular consumers, 2-4TB is likely way more than 95% will ever need. There are a lot of people on these message boards who have built their own 30 TB media server and are constantly running out of space, and therefore would prefer cheap 10+TB drives with good performance, but that's really a very small market. Anyone who needs massive storage space for professional reasons can write it off as a business expense.
I don't think we're going to see huge increases in capacity or huge decreases in $/GB, because most people stream from Netflix and play console games, and don't really need to store large amounts of data.
jabber - Friday, April 28, 2017 - linkI'll take a simpler single platter 1.5GB version please.