A flash drive in your future?

A flash drive in your future?

Summary: The good, the bad and the uglyI've been a huge fan of flash ever since I plunked down $400 for a 10 MB compact flash card in the early '90s for my brand-spanking new HP Omnibook 300. Light and built like a tank, the 300's chiseled abs flash drive almost doubled battery life to over 10 hours.

SHARE:
17

The good, the bad and the ugly I've been a huge fan of flash ever since I plunked down $400 for a 10 MB compact flash card in the early '90s for my brand-spanking new HP Omnibook 300. Light and built like a tank, the 300's chiseled abs flash drive almost doubled battery life to over 10 hours. And the sleep mode actually worked - the only Windows notebook I've used that did.

I was finally forced to retire the 300 after using it daily for over five years due to software bloat. But the warm glow of a really tough, light and reliable notebook - I think it was designed by HP's calculator division - how about an encore, Mark - has kept me hoping for a general return of flash drives.

It looks like my time has come. Almost. Starting later this year.

Despite announcements, few products Flash vendors have been making plenty of announcements, but few show up at Fry's. The OEM announcements are designed to entice notebook vendors, not you and me. With the Sony VAIO G1 flash version, that is finally starting to change.

What to expect? The good:

  • Faster boot times and app and document loads.
  • Fast large-block writes.
  • 2-4 ounces lighter.
  • Wear-leveling works: the fact that SLC flash can only handle a couple of hundred thousand read/write cycles won't be an issue for notebook users as the flash will last as long as hard drives. I've done the math.
  • .

The bad:

  • Only 30-60 minutes more battery life. Why? In today's laptops the drive is less than a quarter of the load, way less if you've got energy saving features turned on. Reducing drive load to zero buys something, but not as much as it did with my Omnibook.
  • Lower capacity than hard drives - largest flash disk is 64 GB.

The ugly:

  • Random write performance is so poor that vendors don't quote numbers. And Windows and Mac OS turn most reads into a read and a write.
  • Cost: even with flash prices dropping 70% a year, flash is still 10-40x disk. It will catch up, but not this decade.

The Storage Bits take Flash drives will usher in a new era of long-life ultra-light notebooks, but only as part of a total system redesign. The rumored MacBook Nano will be such a clean sheet design, using a flash drive, LED backlighting, ultra low voltage Core Solo and the latest low-voltage wireless chip to create an ultralight notebook with well over 6 hour battery life.

Comments welcome, of course.

Topics: Laptops, Hardware, Mobility

About

Robin Harris has been a computer buff for over 35 years and selling and marketing data storage for over 30 years in companies large and small.

Kick off your day with ZDNet's daily email newsletter. It's the freshest tech news and opinion, served hot. Get it.

Talkback

17 comments
Log in or register to join the discussion
  • Any Thoughts on Reliability?

    In a discussion I had a with a friend about these new laptops a few weeks back, he
    very enthusiastically predicted that flash-drive replacements will be "more
    reliable" than hard drives.

    I asked him at the time why he'd make such a prediction, and it came out that he
    basically just "likes" solid-state technology, and he was basing the prediction on
    his personal preferences.

    I'm no expert on flash technology--I've been using it for decades, but I've only
    had a pocket flash-drive for about a year, and haven't had any problems with it.

    Is the data stored in flash as "robust" as data written to HDD?

    Is there any data on how resistant flash is to corruption caused by things like EMF?
    blunderdog
    • Mechanicals?

      I'm guessing the mechanical failure of a HDD is not going to be an issue on a flash drive, and thus the assumption that flash will be much more reliable than HDD. I'll bet mechanical failure is the leading cause of HDD failure, as opposed to failure of the magnetic media of the HDD, or other "soft" failures. Correct me if I'm wrong, fellow posters.
      ejhonda
      • Well it appears most drive failures may be surface related

        Todays drives are using ECC (Error Correcting Code) by default, the re are constant sector read fails, the drives are programmed to allow so many bad reads before moving the sector and marking the current one bad. It use to be you get drives very rarely had to correct for errors, now it is the standard, on "GOOD" drive today it could have 100's of failed (but corrected by ECC) reads a day and still be considered "GOOD". I believe that a lot of the drives that fail to boot can be repaired with a tool called spinrite
        http://grc.com/spinrite
        mrlinux
    • There is less to go wrong

      Disk drives are amazing devices. They are also very complicated electro-
      mechanical machines.

      Motors, plated media, embedded servo, statistical analysis of the signal coming
      from the head, buffers, control logic and much more. A single controller chip
      manages a flash drive. IMHO, flash has to be more reliable than disk drives.

      Also flash fails safely, refusing to write new data but keeping your old data. Disks
      will win on capacity, price and random write performance, not reliability, against
      flash SSDs.

      Robin
      R Harris
  • Measurments

    2-4 ounces means nothing to me. Sure 5 pounds means something and 100 grams means something but clueless on ounces. To me an ounce is shot of liquor so that lap tops like 2-4 shots less in wieght. LOL What screwed up world with half metric have imperial depending which version you use.
    voska
    • you're not far off

      you're not far off...
      [b] like 2-4 shots less weight [/b]
      Yeah, that's close enough.

      "english" units have a practical application to the trade that invented them. It's just that in the modern world, most of us have lost connections to practical trades. They were never supposed to be scientific nor consistent.

      Metric units are scientific, though not 100% consistent. And certainly not practical. Only a few physicists may appreciate how many wavelengths of [argon/neon/krypton] gas emission a meter is.
      impala_sc
  • SSD Performance/Reliability

    See [url=http://www.sandisk.com/Assets/File/OEM/Manuals/SSD_2.5_and_1.8_Family_Brochure.pdf]page 4 of the SanDisk 32G Product Brochure[/url] relative for performance stats.

    You do the math! ;)

    Pricey but slowly coming down.
    D T Schmitz
    • Note what they leave out

      There are no random write stats in the SanDisk brochure.

      My point exactly.
      R Harris
      • StorageMojo No Go!

        [url=http://storagemojo.com/?p=421]StorageMojo[/url] says:

        ..."[b]Page writes are non-trivial[/b]
        An STM page write consists of five steps:

        1. One bus cycle to set up the page write
        2. Four bus cycles to input the write address
        3. Data is then loaded into the page buffer
        4. One bus cycle to command the write/erase/read controller to start
        5. The controller then writes the data into the page

        Typically each page must be written to in order, so to write the last page of a block means writing every other page first.

        Samsung claims about a maximum (in storage vendor lingo, maximum means ?theoretically possible but you?ll never see it in real life?) random write bandwidth of 12 MB/sec. Let?s call it 10 MB/sec write speed on the 32 GB SSD. To write a full 128k block takes about 13 milliseconds. Block erases take from 2-6 milliseconds on the parts I looked at.

        [b]Oh, so that?s why they left out the IOPS[/b]
        I?ve left out considerations of cycle times, wear-leveling and other data transfer operations that occur as a system gets ready to write to a NAND flash SSD. I estimate is that each random write takes 20-25 milliseconds, roughly the speed of a 1.8″ hard drive, or 40-50 random writes per second.

        If you are used to a 7200 RPM drive, the slower write speed will be a noticeable slowdown..."

        Ah Hah! er ah Hmmmm. Bummer. ;)
        D T Schmitz
  • Forget about Flash!

    I thought flash devices as the future as capacity got bigger and bigger. So I bought 2GB flash drive and was using for backups. I often copy 1GB+ data over 1000 files at a time. The consequence was that it got burnt destroying my precious backup files! It happened in within 6 months. Probably I wrote less than 20 times.

    Writting speed is too slow! And writting bulk generates lots of heat and make device very unreliable. Of course, you have the limits in writting times.

    This means that flash will not be the future.
    What a disappointment.
    joemartn
  • Why would random writes be slow? Flash doesn't have moving parts!

    Why would random writes be slow? Flash doesn't have moving parts like heads that have to move around to a new location!

    Writes are direct to a location - Flash doesn't distinguish between random writes and sequential writes because there's actually no difference between the two.

    The only reason why there's a difference between random and sequential writes on hard drives is because sequential writes are done without moving the hard drive's head. Random writes require repositioning the head, and that makes them a lot slower.

    Flash drives don't have heads to move around, so access is always the same whether it's sequential or random.
    CobraA1
    • Writes are just plain slow in general

      Flash memories work in a manner similar to the old EPROMs which you erased with an I.R. lamp. Initially, all bits are zero (for NAND Flash parts). When you write data to the memory location, it is like breaking a wire; easy to break, but not so easy to put back. In order to change the 0 back to a 1, you need to "flash" the location with a high current. In addition to this, most flash parts have an additional limitation that a whole large block of data (sector) must be erased at any time. So if there is data in the same sector you want to keep, you need to move it as well. At the low level, the process is painfully slow; a sector erase can take from 200 uS to 5 seconds in the Spansion (AMD/Fujitsu) part I am using at work for a new hardware design. To mitigate this, you want to perform sector erases ahead of time and manage the memory locations like a disk. Additionally, a single memory location has a finite number of erases it can endure, so the management scheme tries to spread out where the data is stored when it is changed.

      I'm certain there is a good amount of caching and other tricks used for a hard-drive replacement option, but realize there are fundamental limits in the current technology.

      BTW, the large variance on the sector erase time is because the part actually monitors the temperature, sets all bits to zero, then erases, then reads back the data; sometimes twice. If the temperature is not stable it may need to do this many times before the erase is successful. If the temperature is at the parts extremes, it takes much longer.

      As far as the stability of existing written data, I have not seen any specs on this. Just because there are no moving parts doesn't mean electrical charges won't degrade over time.
      Mr. Big
      • Whoops

        Second sentence should say all bits are set to a value of 1 initially (and after a erase)
        Mr. Big
      • RE: Writes are just plain slow in general

        Mr Big,

        Much thanks for your excellent description of how flash works. I had no idea the under-the-hood details were so complex.

        gary
        gdstark13
  • What about Firmware OS?

    Despite Microsoft's desire for us to buy a new OS, I'm just fine with Winblows XP sp2.x. It's relatively stable and hassle free. Add in goodies that come from Mozilla and Open Office and you've got apps that will pick up where you left off if it DOES crash, so:

    How about a firmware OS? That's a big part of the HD, and a big part of the pain in the butt when it crashes, too. Sure, SOME stuff will still need to go on the HD (preferences, security updates (rolls eyes) and the like, but even those could be reflashed on a secondary chip, maybe? Seems we'd be a lot better off with an OS that was built in. Hell, have a few of them with multiple sockets, and you can boot from any of the above. Have a panel in the bottom of the laptop and change 'em out the way we used to change memory from a board, if there's need for a major update.

    What would it save? Maybe not much. Maybe just some HD space, and some hours and hassles with OS reinstalls when a HD crashes. Microsoft might like it because it'd be one way to make sure everyone is buying their OS (yeah as if that could cure it) and it'd load pretty much instantly.

    Call me crazy, but I like the idea. Or if ya DON'T call me crazy, don't applaud, just throw money.

    Peace & Creation,

    JT
    jt@...
    • RE: What about Firmware OS?

      JT,

      I've been asking this same question for years. With a Firmware OS you can physically protect it from virus, spyware, and dumb users. The current approach of issuing patch after patch has the inherent assumption that the goodguys are forever smarter than the badguys, which just isn't the case. I'm a bit bummed to read in this column about how slow flash memory is, but presumably that's something that continues to improve.

      But imagine a computer that CANNOT be altered. And it's always on. No anti-virus, no mindless patching. This is the ONLY way that the PC will ever achieve the noble status of "appliance".

      gary
      gdstark13
  • Rainbow Technology

    Hey Robin,

    Whats your take on Rainbow technology for storing data. Though its still not yet matured, but looks promising isnt it!!

    http://content.msn.co.in/Technology/TechnologyTT_140407_1615.htm

    Regards,
    Deep
    deep@...