Linux myths & truths

Discussion in 'NZ Computing' started by Lawrence D'Oliveiro, Jul 24, 2006.

  1. Here's a report <> on
    long-time Linux kernel guru Greg Kroah-Hartman's closing keynote at the
    Ottawa Linux Symposium. He addressed some oft-repeated claims about the
    state of the kernel:

    "Plug and play is not at the level of Windows." Fact:

    Linux ... supports more devices out of the box than any other operating
    system ever has. Linux is often even ahead of the pack, being the first
    operating system to implement both USB2 and bluetooth.

    "The Kernel needs a stable API or no vendors will make drivers for Linux."

    Linus doesn't want a stable API, he said. The USB stack, for one, has
    been re-implemented three times so far. Linux now has the fastest USB
    stack available, limited only by the hardware. Linux is lean and

    Windows, too, has rewritten the USB stack 3 times, he noted, but all
    three of them have to stick around in the system to support the various
    and uncontrolled old independent drivers kicking around.

    In fact, Kroah-Hartman has said elsewhere that the Linux USB stack is now so
    powerful and efficient, you can write code running entirely in userland (no
    special kernel modules) that will saturate the full bandwidth of USB 2.0.

    Long-time Windows users will know the problem of abandonment by vendors of
    driver support for old devices, where the last-released drivers won't work
    on newer OS versions. That need never happen with Linux--Kroah-Hartman
    mentions pieces of driver code in the kernel which are only known to have
    one or two users at most, yet they continue to be maintained as a standard
    part of the kernel sources. With closed-source drivers, this would simply
    be uneconomic.

    And in fact, it is not legal to distribute closed-source drivers for Linux.
    "Companies that have intellectual property they say they want to protect
    should not use Linux."

    Finally, who is the Borg, Microsoft or Linux? "By size, Microsoft is the
    Borg, but by function, it is Linux."
    Lawrence D'Oliveiro, Jul 24, 2006
    1. Advertisements

  2. Lawrence D'Oliveiro

    Gordon Guest

    Ms Penguin says, who cares, as long as the fish are caught.

    Myself, must have experienced such changes and yet it is still plug and
    lets connect.

    Stable is good for the horse you are riding at the present, we are able to
    switch horses on the far river bank.
    Gordon, Jul 24, 2006
    1. Advertisements

  3. Lawrence D'Oliveiro

    Elsan Shoop Guest

    So where is a Linux distro using the 2.6.25 kernel or better, that will
    correctly detect and natively support, and install on, and be bootable from,
    a RAID array?

    Not Suse 10
    Not Ubuntu Breezy (old kernel) or Dapper (new kernel).

    That's a problem

    elsan shoop
    Elsan Shoop, Jul 25, 2006
  4. Yes.. alot do.. but it depends on the hardware of the Controller of the RAID
    Array. I've used Linux on a lot of hardware and alot of (for example default
    debian install) will find and use the RAID arrays perfectly well for

    Craig Whitmore, Jul 25, 2006
  5. I have used several RAID arrays with Linux kernels back to 2.4, SuSE 8.1 &
    Gentoo 2003-something. Some of them were even booting off RAID.

    In general I'm not keen on hardware RAID with Linux. It works, but not well
    enough for me to be completely happy. These days I would recommend Linux's
    built-in software RAID.
    Lawrence D'Oliveiro, Jul 25, 2006
  6. Lawrence D'Oliveiro

    Steve Guest

    Is that *real* hardware raid, like my adaptec raid boxes at work, or linux software raid
    software raid? All of them work perfectly. You have to tell the installer
    how to set it up... after all, it's not a mindreader, and the options are
    many ( raid 0/1/4/5 at disk level, at partition level, lvm, ... )

    And, of course, I'll let you know about the kernel when we get there. Last
    time I looked we were only up to, so 2.6.25's a while off yet.

    My attitude is that if you want raid, you're building a server, so you
    need to actually know (or learn) what you're doing first.

    Steve, Jul 25, 2006
  7. Lawrence D'Oliveiro

    thingy Guest

    No way in raid is no where as good as hardware raid...a
    few bad crashes and loss of data on software raid means hell freezing
    over beofre I uses it on serious applications....

    In the same period I have used lots of hardware raid without an issue.


    thingy, Jul 26, 2006
  8. Lawrence D'Oliveiro

    thingy Guest

    Depends on the controller used, some hardware arrays are harder to get
    going on than others. The biggest problems I have had is with irq or
    iomem conflict stopping the hardware detection and then you see no disks.

    Otherwise even booting off a SAN is easy, as long as the LUN number
    <256.....but not many people have over 256 Luns.....


    thingy, Jul 26, 2006
  9. The hardware RAID drivers may work, but what kind of admin tools do you
    have? RAID is not worth using unless you have some ability to check the
    health of your drives, add/remove drives etc. Most of the hardware RAID
    systems I've used under Linux had deficiencies in this respect. Whereas
    with software RAID, it's a complete and very capable solution.
    Lawrence D'Oliveiro, Jul 27, 2006
  10. Lawrence D'Oliveiro

    Don Stokes Guest

    Software RAID, IME, only works nicely if your hardware is happy to have
    things fall down and go boom under it. I've seen software RAID just
    lock up completely when drives fail -- sure, you get to reboot onto a
    workable remaining disk, but it's still a dead box until you do.

    Haven't tried, but I suspect SATA controllers are better at handling
    dying disks than crummy onboard IDE controllers. SCSI is usually fine,
    but horrendously expensive on a price per gig basis.

    Personally, I've had pretty good results from the 3ware IDE RAID
    controllers and cheaparse tricky-dicks removable IDE trays. The Linux
    management utility is a wee bit clunky, but I've dealt with worse.

    -- don
    Don Stokes, Jul 27, 2006
  11. For my current clients, data integrity tends to take precedence over uptime.
    That's why I now insist on proper admin tools before using any RAID
    solution. If I want better than four-nines uptime, I'd put in additional
    redundant servers.

    Besides, software RAID opens the door to other interesting possibilities,
    like DRBD.
    Lawrence D'Oliveiro, Jul 28, 2006
  12. He's posted his full presentation online at
    <>. He lists over half
    a dozen technologies that the Linux kernel was the first in the world to
    support. And pointing out an advantage of open-source drivers:

    And remember, almost every different driver that we support, runs on
    every one of those different platforms. This is something that no one
    else has ever done in the history of computing. It's just amazing at how
    flexible and how powerful Linux is this way.

    He also gives a reason why the kernel developers are happy to accept new
    drivers into the kernel source tree, no matter how obscure: because it
    gives them more samples from which to see patterns, from which they can
    extract the commonality, turn it into reusable common code, and reduce the
    amount of work needed to write the drivers. He gives the example of two
    different drivers being written for two different USB data-acquisition
    devices; because they were open-source, it could be seen that the hardware
    was so similar they could be merged into one driver, with a concomitant
    reduction in maintenance. With closed source, this kind of factoring never
    could have happened.

    The review process for kernel code submissions, as informal as it may be,
    can sometimes seem harsh. Though it doesn't seem that different from trying
    to conduct a discussion on USENET, really. Example comment from Greg-KH

    Wow, for such a small file, every single function was incorrect. And
    you abused sysfs in a new and inter[e]sting way that I didn't think was
    even possible. I think this is two new records you have set here,

    (By the way, the LKML posting this comes from is archived here

    For those who want to help with kernel testing, and happen to have a spare
    machine or two lying around to use for crash fodder, an invitation:

    And if you really feel brave, please, run Andrew Morton's -mm kernel
    tree. It contains all of the different kernel maintainer's development
    trees combined into one big mass of instability. It is the proving
    ground for what will eventually go into Linus's kernel tree. So we need
    people testing this kernel out to report problems early, before they go
    into Linus's tree.

    In conclusion: "Total world domination is proceeding as planned."
    Lawrence D'Oliveiro, Jul 28, 2006
    1. Advertisements

Ask a Question

Want to reply to this thread or ask your own question?

You'll need to choose a username for the site, which only take a couple of moments (here). After that, you can post your question and our members will help you out.