back to article Toshiba gets NASty: Soups up hi-cap enterprise drive

Toshiba has developed a high-capacity NAS drive from its 10TB and 14TB helium-filled enterprise capacity spinners. The MN07, derived from the MG07ACA, comes in 10 and 14TB capacities and Toshiba has said it has 9 platters, meaning 1.555TB/platter at 14TB, and is suited for NAS enclosures with 8 or more 3.5-inch bays. …

  1. LeahroyNake

    Hey Mamr

    'and microowave assisted magnetic recording (MAMR)'

    As cool as microowaves sound I'm more interested in the part about NAS with more than 8 bays ? If I could afford one let alone two they would be really nice in a Qnap 2 bay!

    Also load/unload cycle, is that the same as power cycles ?

    1. vincent himpe

      Re: Hey Mamr

      Load/ Unload means head loading onto the platters and retracting them off into the parking garage.

      Modern drives no longer land the heads on the platters on powerdown or idle. The heads are retracted and slide into a 'garage'. This is technically a misnomer. What happens is the arms pull back and slide onto a ramp. the lifts the heads up and away from the platter surfaces. The arm then lands in a notch that holds it. The heads are dangling free.

      A headstack 'parked' this way is very resilient towards damage and shock. There is no risk of the heads banging into anything and chipping the ferromagnetic surfaces.

      Whenever a drive is not used for active data transfer a timeout system will command the headstack to retract. The motor keeps spinning so the platters remain up to speed although they can 'float' a few thousand RPM. They will stay above a certain minimum so the heads can be 'loaded' at will.

      A minimum RPM of the platters is required to create enough surface pressure to keep the heads floating and not slam into the surface.

      Another common misbelief is that people think the heads 'crash' into the platters and cause damage. A head crash damages the HEAD. The surface of a harddisk is so hard (made from diamond-like carbon deposit. it is actually a carbon layer grown in a diamond lattice) you cannot scratch this. the magnetic layer is buried under that surface. the top of the carbon layer is lubricated with a coating to increase surface tension. This repels the heads away from the surface.

      The spinning platters create a suction effect by pulling air between the platter surface and the head. It is basically an airplane wing upside down. The heads contain a heater element. By altering the temperature of the head they can control the flying height. As the head is writing it also heats up and the pole tips extend. By altering the heater temperature they can compensate for this. Control algorithms seek for maximum signal strength and not only drive the head towards the center of a track but also seek in vertical space for optimum flying height.

      I worked for years in the HDD world and it still amazes me how they can cram all that data on such small a surface with this reliability. Some of the stuff they do is like Evel Knievel doing motorcycle stunts...

      Prediction algorithms figure out what will pass at what moment in time underneath the head and steer the headstack to the right spot , in time for the data sector to fly by. the head accelerates, maintains speed , then decelerates just in time to pick up the 'pilot' that precedes the data block. The pilot is phase locked to fine control drive rotational speed. depending on the track we are on we know exactly how many microseconds of data will pass by before we will hit another pilot block. The drive times all that stuff out and then turns on the writer for that precise amount of time to spit the data onto the disk.

      While writing a block the drive is 'blind'. It all depends on accurate timing so not to write beyond track end, and accurate positioning , so not to write into adjacent tracks. It's like doing a run around an athletic course, you have to run 100m in 10 seconds , staying in your assigned track . stopping just short of the finish line ( stepping on the finish line corrupts the pilot, going beyond throws data in the next block corrupting that) , all while blindfolded.

      1. Piro Silver badge

        Re: Hey Mamr

        I'd just like to say thanks for this quality post.

        1. Lord Elpuss Silver badge

          Re: Hey Mamr

          Me too. Fascinating.

      2. LeahroyNake

        Re: Hey Mamr

        @Vincent thanks for the awesome post, I think it's worthy of a full Reg article ! Very well written.

  2. Anonymous Coward
    Anonymous Coward

    Price Fixing?

    Any hard drive 5TB and above, appear price fixed, yet CMA do nothing.

  3. expat_abroad

    Hard disks stuck at 3.5" - why?

    One thing I simply don't get is why HDD manufacturers are still sticking to the 3.5" disk format. I mean, let's face it - spinning rust lost the speed race to SSDs years ago. It's main and key advantage is capacity/price, which SSDs simply can't touch for now. I don't buy drives like these 10 TB here for speed, but for lot's of cheap space, and I can always use more. So, why not go back to, let's say, 5.25 inch format platters, which offer an ~ 80% larger area (3.5 inch disks are actually a bit over 3.7 inches, and 5.25 disks are about 5.1 inch in diameter). These larger disks fell into disfavor due to higher loads on the motors and spindles and worse thermals many years ago, BUT, that was long before we had current disk materials and He-filled drives that enable thinner, lighter disks and reduced friction. Plus, current bays for 3.5 inch drives (except laptops) are 5.25 inch bays, after all, so they would fit in my existing racks or NASs. So, Toshiba, WD, Seagate & Co., how about it? Instead of a 10 TB capacity, we'd instantly go to 18 TB, sticking to PMR. Thoughts, Comments?

    1. vincent himpe

      Re: Hard disks stuck at 3.5" - why?

      Torque and access speed.

      spinning such massive platters at high RPM requires a lot of torque..

      a larger diameter also means more travel time for the headstack making the driver slower. jumping from the outward to the inward track simply takes longer on a 5 1/4 vs a 3 1/2 drive.

      if you plot spinspeed and travel time you will find one sweet spot for drive diameter vs access time.

      thats why we use 3 1/2. it falls slap dab in the middle of that range

  4. Palladium

    According to Amazon, USD550 for those packrats who value their TB per physical space over TB/$.

  5. Roland 2

    A little surprised by that bit:

    "The MNO7 has workload ratings of up to 180TB per year. "

    It being a 14TB drive, does this mean the endurance is 0,03 full drive write per day ?

    I thought endurance was still a selling point vs. SSDs.

POST COMMENT House rules

Not a member of The Register? Create a new account here.

  • Enter your comment

  • Add an icon

Anonymous cowards cannot choose their icon

Other stories you might like