SoylentNews is people
SoylentNews
Toshiba is sampling a 9-platter, 14 terabyte hard disk drive that uses "conventional magnetic recording", aka the traditional perpendicular magnetic recording (PMR) with no shingling:
The new series comes with both 14TB and 12TB disks that wield nine and eight platters, respectively. Toshiba also becomes the only company with a nine-platter drive with 18 heads. Each platter packs 1.56TB of data storage.
Competing HDD vendors (WD and Seagate) have used helium designs for several years, so Toshiba has largely been considered late to adopting a helium design. Toshiba fills the 3.5" drives with helium instead of air and uses a laser sealing process to contain the gas. The helium reduces internal air turbulence from the spinning disk. In turn, it reduces vibration and provides power, performance, and reliability advantages. It also allows the company to use thinner platters, which facilitates the additional ninth platter.
While Toshiba may be the last HDD vendor to market with a helium HDD, the company did it in style. The MG078ACA, which carries a tongue-twisting name because it is destined for the data center, currently weighs in as the densest HDD on the market using conventional recording techniques. That represents a 40% increase in density over Toshiba's previous-gen 10TB models.
[...] Toshiba currently has 24% of the HDD market share according to Coughlin and Associates, which comes in third to Seagate (36%) and Western Digital (40%). The company has been surprisingly resilient and has clawed back market share over the last year. The addition of a class-leading 14TB model should help it gain even more market share over the coming year.
Both drives have a 5 year warranty.
1.8 TB 9th-generation PMR platters are possible and could be used in a 16 TB Toshiba HDD late next year. Will we see 2 TB per platter without the use of HAMR/MAMR or shingles? Combine that with 12 platters (using a glass substrate), and suddenly you can have a 24 TB HDD.
Also at AnandTech. Previous article.
Previously: Western Digital Announces 12-14 TB Hard Drives and an 8 TB SSD
Seagate's 12 TB HDDs Are in Use, and 16 TB is Planned for 2018
Glass Substrate Could Enable Hard Drives With 12 Platters
Seagate Launches Consumer-Oriented 12 TB Drives
Western Digital to Use Microwave Assisted Magnetic Recording to Produce 40 TB HDDs by 2025
Western Digital Shipping 14 TB Helium-Filled Shingled Magnetic Recording Hard Drives
Seagate to Stay the Course With HAMR HDDs, Plans 20 TB by 2020, ~50 TB Before 2025
(Score: 1, Interesting) by Anonymous Coward on Saturday December 09 2017, @08:54AM (9 children)
I remember when hard drives used to be physically bigger, to go in a so-called 5.25-inch bay. The last was the Quantum Bigfoot. Before that some of them were "full height," the height of the floppy drives in the original IBM PC, which was like the height of two modern CD/DVD drives. In those days, if you had a 10 MB or 20 MB drive you were doing all right. With today's densities, a drive of that size could hold tens of terabytes.
(Score: 4, Insightful) by Unixnut on Saturday December 09 2017, @09:48AM
Yes, I remember those too, although never had the pleasure of seeing a bigfoot drive myself (though I still have a couple of 4GB Quantum fireballs whirring away in a machine, those things just don't die). Always seemed like an elegant solution for the home PC because most cases had at least 2 5.25" slots, and usually only one was ever in use.
The problem of using bigger drives now is that many of these drives go into rackmount servers for "Cloud storage", "Big data" etc... and none of those systems have 5.25 inch bays, some don't even have 3.5", relying on 2.5" disks.
On the non enterprise side of things (which were the bigfoot drives original market), most people I know don't even have desktops, if they have a PC it is a laptop, but quite a few get by on nothing more than their phone and tablets. Some have an external 2.5" 1TB drive, or 3.5" 6TB drive for backups/media storage, and they are quite happy with the current sizes for their needs.
From a enterprise point of view, having one bigger drive with more capacity vs 3 smaller drives in some sort of RAID makes little sense. Not to mention making the platters larger will increase seek times, and also bring down the max rpm of the spindle (because the edges of the platters will be spinning much faster than the centre for a given speed compared to a smaller drive).
To retool everything for 5.25" drives (or the taller "full height" 3.5's of old) and basically break compatibility with existing infrastructure just to pack more data in a "disk drive" isn't worth it, especially as you don't get the redundancy of multiple disks and increased costs due to reduced economies of scale.
(Score: 2) by damnbunni on Saturday December 09 2017, @09:57AM (1 child)
The problem with the Quantum Bigfoot drives was that they were SLOW. Very slow. Hideously slow.
I had a couple of them. Sequential transfer speeds were 'okay', but seek times were horrible and random access was a snorefest.
(Score: 1) by toddestan on Sunday December 10 2017, @06:37AM
I'd still like one as cheap, large, and doesn't have to be particularly fast mass storage drive. There's no reason to use one for a boot/OS drive. We have SSDs for that.
(Score: 2) by c0lo on Saturday December 09 2017, @10:20AM (3 children)
I remember when the harddisks had swappable plates [auckland.ac.nz].
https://www.youtube.com/watch?v=aoFiw2jMy-0 https://soylentnews.org/~MichaelDavidCrawford
(Score: 0) by Anonymous Coward on Saturday December 09 2017, @05:22PM
Bernolli(sp?) boxes.
(Score: 2) by Unixnut on Tuesday December 12 2017, @08:35AM (1 child)
You know you are hardcore when you need a socket wrench to switch your hard disk platters :-)
Very cool, thanks for the share!
(Score: 2) by c0lo on Tuesday December 12 2017, @09:37AM
I actually tried to run programs on a machine using those (uni time)
Input from punch cards, compile errors (or any output) on listing. After 2-3 cycles of 'Here's the Fortran program on this stack of cards' - one week later 'Here's the compile-error listing, go repunch the typos', I gave up and started to use a ZX Spectrum clone.
(The above just in case you acquire the taste for retrocomputing)
https://www.youtube.com/watch?v=aoFiw2jMy-0 https://soylentnews.org/~MichaelDavidCrawford
(Score: 2) by takyon on Saturday December 09 2017, @02:00PM (1 child)
Vibrations and speed would be a problem. But it would sure be something to see a helium-filled 5.25" drive (which should be able to spin faster).
[SIG] 10/28/2017: Soylent Upgrade v14 [soylentnews.org]
(Score: 0) by Anonymous Coward on Sunday December 10 2017, @12:36AM
why would vibration and speed be a problem with the same physical size and enclosure?
Those full height drives often were 3600rpm -- they were not tiny spinners. 15krpm drives were certainly not the class of drives those old 5.25" full height MFM/RLL and SCSI drives were capable of.
Seek times will always be a problem on a platter that wide. Think of records or laser discs. Even with a giant shiny disc read with a whopping laser... skipping chapters was about as time consuming as getting up to move the needle on the record, except with laser discs one merely needed to wait and marvel at how convenient technology had become while you waited.
CD-i, and then DVDs greatly reduced that delay and latency, often using buffering and caches to read ahead the table of contents and know right where to go. Data CDs were usually dependent on the speed the drive mechansim could spin and for the laser to track; then every time it'd have to try to figure out where stuff was. Some music players traced over the disk to find that out, but over time that info got recorded onto the disk as a header (I have to admit I am not entirely familiar with that part of ancient history).
anyway I set up a raid 5 of full height 3.5" scsi-2 wide drives; 30GB. For the era (late 90s/early aughts) read was fast, seek was not...