December 18, 2005

Diskless PC possibilities

I’m not a hardware guy, so please pardon me if some specifics here are implausible, but an interesting idea has arisen, and indeed turned into the subject of my December Computerworld column. Shayne Nelson raised the subject of diskless PCs based on USB/flash “drives,” and a web search uncovered a Slashdot discussion on the subject a couple of months earlier, which in turn seems to have been based on a now unavailable Yahoo story. The technology certainly would seem to be practical in the near future, and it raises some interesting ramifications and possibilities.

1. The most vulnerable, volatile, and valuable parts of the computer — the programs and data — could now be removable and put in any pocket, or mailed without much fear of breakage. Even better, they could be segmented, on multiple drives per computer. Possible security and administration benefits include:

a. The drive that stores most programs could be locked down, tight. Pick your dream technology or policy for making PC images consistent across your network; it just became a lot more plausible to implement.

b. The drive that stores most data could be entirely encrypted. Flash drive access is several orders of magnitude faster than disk access, making this a reasonable precaution even though it’s not very practical with magnetic storage.

c. What’s more, laptops might still be lost or stolen — but they wouldn’t have to have data on them! An employee whose laptop is stolen is unlucky. An employee who leaves sensitive data in an unattended laptop could now be justifiably fired.

d. In two-factor authentication, the flash drive might be the second factor. No fuss, no bother.

e. You could physically upgrade every user’s disk without shipping PCs around. Just ship flash drives around instead.

f. Enterprises could implement a policy of NO PERSONAL WEB SURFING UNLESS YOU SWAP OUT COMPANY DRIVES (and therefore presumably swap in your personal ones). All kinds of security problems would be ameliorated ASAP, at much less cost to employee goodwill than more draconian crackdowns incur.

2. Environment-specific computer equipment would now be much more affordable. Classrooms, meeting rooms, operating rooms, etc. might have more suitable devices than they now do.

3. Before long, we might not need to travel with laptop computers! Yes, devices in hotel rooms might be problematic from a security standpoint, but there are workarounds for that too. And in any environment that’s more locked down, such as a home or corporate office, the problem is almost nonexistent unless you work for a Three Letter Agency.

4. Disk space would initially be decreased, just as the Web initially made UIs worse. That’s not a huge problem, but it might not bode well for bloatware vendors (e.g., Microsoft).

Obviously, this isn’t a big deal from a business standpoint until the devices are actually manufactured and sold. But it’s fun to think about. And it actually makes a whole lot of sense.


14 Responses to “Diskless PC possibilities”

  1. Brian Carter on December 19th, 2005 6:10 pm

    Hi Curt:

    Ardence streams both the OS and apps on-demand to servers/desktops with
    or WITHOUT hard drives…Deployed throughout the national nuclear weapons
    complex and currently being deployed on a Dell diskless SmartClient for the
    USAF in Europe.

    The following appered in ComputerWorld

    Application streaming creates a virtualized desktop that can be managed
    centrally, yet offers the speed of local execution.

    News Story by Robert L. Mitchell

    NOVEMBER 21, 2005 (COMPUTERWORLD) – Automated software distribution has been a
    hot topic in desktop management, but the next big thing is on-demand software
    delivery. While ASD tools help control desktop support costs by making software
    installations consistent, the on-demand software-delivery technologies go one
    step further: They can virtualize the local installation and stream the
    applications — and even the operating system — from a central distribution
    server in real time.

    Sanjeev Shetty, director of IT at Time Warner Cable in Greensboro, N.C., is using streaming technology to manage the desktop application environments in the company’s
    300-seat call center. Shetty says he considered thin clients but couldn’t
    justify the back-end server farm investment required to support an architecture
    for Microsoft Terminal Services or Citrix Systems Inc.’s Presentation Server
    (previously MetaFrame).

    Instead, he installed Ardence Desktop from Ardence Inc. in Waltham, Mass.
    It creates and stores complete system images on a server and streams portions of the operating system and applications to desktop users at boot-up. “It didn’t require a large investment in server infrastructure and provided immediate ROI,” Shetty says.

  2. DBMS2 — DataBase Management System Services»Blog Archive » Solid state (Flash) memory vs. RAM vs. disks on December 20th, 2005 12:20 am

    [...] I just wrote a column and a blog post on the potential for diskless PCs based on flash drives. It was a fun exercise, and I think I kept it general enough that my lack of knowledge about hardware technology details didn’t lead me into significant error. [...]

  3. Richard Veith on December 22nd, 2005 10:54 am

    I too have been intrigued by the prospect of massive amounts of flash memory
    available in very small packages. In the same way that the PC was a major shift
    from dumb terminals, I think diskless PCs or something like that will be a
    major shift from the current PC environment. My take on it, though, was a little
    different in that I thought an iPod-like arrangement (with built-in viewing or
    playing capability) might be the rule. I wrote a rather academic article on it,
    linking the idea to Vannevar Bush’s 1945 Memex concept. The article is still in
    the process of being published, but here is a link to the preprint:

  4. John Thiels on January 2nd, 2006 5:17 am

    Hi Curt,
    Interesting article in Computer World. I am considering using flash memory
    as an experiment in learning Linux for my personal use as a grad student
    writing up a dissertation.

    There are small-scale Linux users already trying out the diskless PC–using
    “thin clients” with added Flash or CF memory for personal use–check out and for models in use by some in those
    communities–puppylinux is quite friendly for non-geeks and you can see some
    of the possibilities for “everyday” PC users and small businesses.

  5. VV on January 15th, 2006 7:02 pm

    Very interesting but, obviously, you do not mention some important things:
    To be useful, data must generally be processed by specific applications (for instance .DOC files are to be processed by MS Word, .JPG files are to be processed by a jpeg capable application, especially if you want to modify said files).
    So, you should not only carry your data with you but also the applications to process them. And these applications must usually be installed into an OS and then they are linked somewhat to the OS. Thus you would need to carry with you: your data, the applications to process them and the OS in which said applications are installed.
    But modern OSes are usually linked to the hardware and, even if the box you use is called a “PC Compatible” computer, this compatility refers to basic (and rather old) “standards”.
    Nowadays, you cannot natively take an existing HDD running Windows XP, move it to another hardware than the one it was installed on and boot this other hardware. If you do so, you will certainly experiment BSoD (Blue Screens of Death). Linux is more permissive but you can’t be sure that a Linux OS installed on some specific hardware will be able to operate all the components of another hardware platform (especially network cards, graphic adapters, hard disk controllers…)

    Some interesting technologies already exist in order to make an existing installed OS compatible with several hardware.
    Neoware for instance uses a technology named “UbiBoot”, that was developed by Qualystem (Neoware acquired Qualystem in 2005). UbiBoot can make an OS installed for a specific hardware platform compatible with another hardware platform, but a detection phase has to be performed before the new hardware platform can actually be operated by the pre-existing OS (installed on the previous HW platform), and sometimes you have to manually adjust things on the existing OS in order to make it able to boot on the second platform.
    The detection phase often requires the user to provide the drivers for the hardware that has just been detected. And this may not be very convenient to the non-technical user (and even the technical user would have to make sure the needed drivers are available, on CDs, flash disks, floppies etc…)

    Neoware uses UbiBoot in conjunction with its Image Manager software suite which, like Ardence product mentioned in an earlier comment, can stream OS, apps and user data to diskless PCs, over a network link.

    But obvioulsy, UbiBoot technology could be used in order to make a 1GB or 2GB flash disk containing OS, apps and data compatible with any kind of PC. Yet the drivers hassle would still exist in many cases. And MS would need to support boot over USB (they have taken late initiatives towards this technology:
    Interesting to note that MS does not mention that their USB drivers stack cannot completely be loaded and initialized in early boot phase. Otherwise, boot over USB would be possible in Windows 2K/XP, as it is under DOS – with appropriate BIOS/int13h support for USBtoIDE- and in Linux)

    Another way to go would be to have hardware platforms that would be 100% compatible or standardized (remember the first IBM PCs and “compatible”?). But this would only be possible if the user accepted lower performances because standards are limitating factors when performances improvements are concerned…

    In my opinion, the autonomous devices will still be the preferred
    way for personnal computing for the next decade. The user will still use PDAs, Laptops, Desktops, and also ThinClients and diskless PCs with shared virtual HDD. Of course, the users will have their data (some of them) on USB keys (backups maybe…)

    One day, maybe, we will only carry our data with us and most of the devices we will be able to use will be able to open our files. And my shoulder will be happy not to have to carry a case with my laptop, various power supplies, a mic and headphones, two set of CDs, various cables etc.

  6. N.B. Weersing on February 21st, 2006 4:10 pm

    From the viewpoint of a systems administrator concerned about simplifying support to users and about compliance with company policies (re: software licensing, data integrity, “safe web-surfing”, etc.), I find many attractive points in the design of a diskless PC that boots from a flash-RAM-stick.
    After deploying such a computing platform throughout a user group, the IS department could enforce company policies that now are often ignored or consciously abused. The company’s standard software build could signal the firewall that only web-sites on an approved list are to be accessed. It could also be configured to prevent the addition of personally owned or downloaded software.

    Calls to the Help Desk for help with software issues could be resolved simply by replacing or refreshing the standard flash-RAM-stick and copying down the configuration files, peculiar to that user, from a central archive.

    I can’t swallow the argument made above that only MS Word can be used to open & update a *.DOC file ( & by inference, that only MS Excel can maintain *.XLW & *.XLS files.) For over five years I’ve been using Star Office (on NT, Linux, OS/2 or SunOS), to update files that I download at home from my office server. (On some machines at home, that suite has been replaced by Open Office.) Under UNIX/Linux, I have several programs that are very adequate for processing image files, whether the format may be JPEG, GIF, TIFF, etc.

    As to dependence on the combination of hardware and the OS, only those programs written in Java, perl, PHP, Python or some other truly portable script, can be readily moved from one platform to another. I feel lucky to have all versions of Star Office 5.1, (for four diverse platforms), fit on a single CD. With subsequent releases, I think a DVD is required, in order to keep the same degree of compact consolidation and portability. With SW bloat having been blessed by Microsoft, I feel that the mindset of competing developers has too often shifted towards more tolerance for redundant or superfluous code. We are very lucky to have the examples of cell ‘phone and PDA software developers to remind us about more sparse, yet effective alternatives.

    [I was saddened to see GeoWorks software manoeuvered out of the PC market by MS machinations, in the early '90's, and then delighted to see its rebirth in the PDA arena, as "Palm OS". Why was this example of abusive monopoly practice not brought up in the anti-trust trial ? To get DR DOS or OS/2 or GeoWorks to co-habit effectively with MS Windows 3.1, I had to clobber krnl386 under Windows. There's a clear example of deliberately predatory software.]

    Thank you for your thought-provoking column.
    — nbw

  7. Jack Hsu on March 9th, 2006 6:00 am

    I think if we still use the usb as the boot device, the pc can not be called
    diskless pc, actually, it still has a disk, but just smaller, data or operating
    system still on there.
    A diskless pc should really has no any storage on it, no disk, no flash, no usb
    device. How to boot it? Network! if we can just boot the pc from the network, all
    the management and software are from the other side of the network, the marketing
    of a diskless pc will come.
    What technology to make a diskess pc to boot from network? as my knowledge, three
    successfully products are out to support boot from network. We already sell one
    of them for one year. The technology uses the pxe and iscsi technogoy, to boot a
    pc via onboard PXE bootrom, the booting code initiates a iSCSI initiator by software
    to mapping a remote hard disk to local computer, then what as usuall, the OSs
    are booting, we now can boot windows 2000/xp/2003 and linux. The drawback of our product
    now is only supported under LAN.

    I saw a product last week, a company designed a card combining the iscsi and lan
    card, namely, by equiping such interface card, a pc will boot from network.
    We also trigger by such product, so we now are developing some kind of bootrom
    code to integrate iscsi initiator and tcp/ip stack into the bootrom, therefore,
    we can achieve booting a pc from WAN.

  8. Curt Monash on March 14th, 2006 5:43 am

    VV — you are correct that I was glossing over issues of hardware compatibility.

    On the other hand, the whole idea only works if hardware “shells” are built expecting plug-and-play flash drives. So rigorous adherence to (possibly new) API/driver standards could be reasonably demanded as a design requirement.


  9. Curt Monash on March 14th, 2006 5:46 am

    Jack — if there isn’t magnetic storage spinning around really fast, I’m content to call it “diskless.” The real point is that USB storage is in a form factor that is physically very portable, whereas disks aren’t.

    The challenge is to also make the storage logically portable …

  10. The Monash Report»Blog Archive » Path to the diskless PC? on March 15th, 2006 3:28 am

    [...] Jimmy Daniels of RealTechNews writes about Robson Flash Cache technology, as an add-on to conventional PCs. But at the bottom he gets to what I think the core point, which is that it would be better to use flash to replace hard drives altogether. [...]

  11. The Monash Report»Blog Archive » The laptop security nightmare on March 16th, 2006 2:33 am

    [...] Yes, I’m repeating myself. • • • [...]

  12. jack on March 18th, 2006 10:37 am

    Mmm…. what the definition os a disk? By using the current technology, disk is
    a media to store data. In other words, rotating/managtic disk or usb storage
    are all called “disk”, I think in the near feature, small size/high capacity storage
    will be popular and cheap. We think the discuss should focus on the next generation
    of the computing! we stongly think “a diskless pc” is the next star of the IT
    industry. The computing model can be divided into 3 eras. The first eva was
    between 1960-1980, the central management and central computing archtitecture,
    all the managenment and computing power are on the server, called mainframe. The
    second era is the pc era. Distributed computing and distributed managment.
    becuase pc becomes so cheap quickly more than the HUGE company thought. so DEC
    was dead soon after the PC grows. What is the next generation computing model,
    it should has the local computing power but having the central managemnet mechanism.
    This diskless pc is the solution. Think! we just buy a pc without any storage,
    just plug the network line into socket in your home, all the operating system and
    software bring to you within minutes, like the TV. We are making the solution component
    now. we think we can change the world. the third era of computing revolution is coming
    and all the data and reports believe it will come in 2007.
    For the enterpise, imaging that, if your pc is broken, just replace a new one.
    and turn on the pc, your familar os and app still on there, how to achieve it ?
    diskless pc, we already achieve it. but now we invest all our resource to make it
    boot from wan….,

  13. The Monash Report»Blog Archive » Flash drives as hard-drive replacements on January 5th, 2007 1:13 am

    [...] But to me, the really interesting future here is PCs with removable persistent solid-state storage. I wrote about the subject a year ago, and I just want to take this opportunity to remind people that’s it’s a desirable and not-implausible way for personal computing and consumer electronics to evolve. If the storage part of the system can be separated out, what you’re left with is mainly the human-facing I/O and the processing power to drive that. So from where I sit, portable external storage could drive an explosion in interesting and useful electronic device form factors. • • • [...]

  14. Ted on December 22nd, 2007 1:11 pm


    This sure as heck beats reading Playboy in the dark wth a flashlight….

Leave a Reply

Feed including blog about enterprise technology strategy and public policy Subscribe to the Monash Research feed via RSS or email:


Search our blogs and white papers

Monash Research blogs

User consulting

Building a short list? Refining your strategic plan? We can help.

Vendor advisory

We tell vendors what's happening -- and, more important, what they should do about it.

Monash Research highlights

Learn about white papers, webcasts, and blog highlights, by RSS or email.