Re: An Open (and very general) Letter To All Companies Contemplating A Switchover

• Previous message: George Jones IV: "Re: Sony Considers Linux for PS3 Hard Disk"
• In reply to: essteeaenn_at_worldbadminton.com: "Re: An Open Letter To The Linux Enthusiasts."
• Next in thread: Peter Köhlmann: "Re: An Open Letter To The Linux Enthusiasts."
• Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]

Date: Fri, 10 Jun 2005 17:00:03 GMT

Note subject change and followups.

In comp.os.linux.advocacy, essteeaenn@worldbadminton.com
<essteeaenn@worldbadminton.com>
 wrote
on Fri, 10 Jun 2005 13:59:49 +0000 (UTC)
<1118411989.717911@cswreg.cos.agilent.com>:
> In comp.os.linux.misc Peter K??hlmann <Peter.Koehlmann@t-online.de> wrote:
>: begin virus.txt.scr John O wrote:
>
>:>>
>:>> I think this is just one of many problems starting out with Windows.
>:> Another
>:>> is Unzipping files. I have to load an application for that too. What
>:> about
>:>> creating MP3 files or burning CDs? Another application, right?
>:>
>:> Wrong. Win XP comes with all this stuff.
>:>
>:> -John O
>
>: Nope
>: It /might/ be bundled with it, but XP just by itself is unable to do that
>
>
> Umm- you _do_ realize that _Linux_ by itself can't even open a window or
> play a DVD or lots of other basic things, right? Those are all part of
> a bundle added to Linux and called a "distribution". Sorry- you'r argument
> holds no water at all.

    panic: No init found. Try passing init= option to kernel.

That's the message generated by line 607 of /usr/src/linux/init/main.c,
and the most probable message one will see if the kernel can even
get past the VFS setup, which I'd have to find. (This is a 2.4.25
gentoo-patched variant. However, I doubt this part of the code has
changed significantly for awhile.)

Linux qua Linux is highly cripped. Not that MS-DOS or
NTOSKRNL.EXE or Unix's or even VMS's kernel would be
much better off; MS-DOS in particular will prompt for
its COMMAND.COM, and NTOSKRNL.EXE might BSOD or simply
sit there (I don't know offhand). There's a few issues
with modules but overall one can't do much with just the
bare kernel.

A fair comparison of a Linux distro to Windows might be
a Linux distro with whatever freeware one likes, compared
with a Windows distro with watever freeware *or* payware
one likes, consistent with budgetary requirements of the
comparison -- probably around $1000-$5000 for a $200 base
hardware computer system. This might include anything
from WinZip to Office, Visual Studio and Visual SourceSafe.

(Yes, there is payware for Linux. One can include that as well,
if one needs to.)

Of course one must ask the obvious question as to whether
one would want to pay all that much, or go with the
Linux distro, which is of similar if not better quality.
However, the comparison must include:

[1] Standardization issues, including de facto. If files are stored
    in Microsoft Word format, one must include time and materials
    in Linux for decoding that format. Fortunately, several options
    for that are readily available, among them OpenOffice, but there's
    even a Java variant, POI:

        http://jakarta.apache.org/poi/index.html

    Of course one might be better off taking a one-time hit and
    convering those documents to something more standard, like
    PDF, HTML, or even XML/RDF. Storing files in MS Word format
    has already been shown to be slightly dangerous, in light
    of the Word95->Word97 fiasco.

[2] Retraining. This is admittedly a given, and for the good people
    not that much of a problem.

[3] Retraining development. This is another given, but is far fuzzier
    than #2; fortunately, it's somewhat less costly as well. Basically,
    someone has to develop #2 in the context of what a worker at one's
    employer is likely to encounter.

[4] Software maintenance. Yes, Linux has it; the costs may be monetary
    (yearly RedHat Enterprise licensing fees), time (searching for and
    downloading the relevant bits), energy (compiling [*] the relevant
    bits), and more time (configuration).

[5] Hardware maintenance. Ideally, this would be equal, but one
    can never be quite sure; Linux works the system differently
    than Windows. For old high-end server applications Linux might
    be slightly cheaper, as it knows how to do HALT (Win95 did not).
    Also, Linux is easily maintainable from a remote location,
    which may require less staff -- or more coordination, to ensure
    everyone's on the "same page".

[6] Bandwidth costs. Ideally, this would be equal as well, but I
    have my doubts about Windows. This may or may not include
    bandwidth generated by malware/rogueware, depending on
    the comparison structure one wants.

[7] Rogueware defense/elimination. I'm not sure even where to start
    on this one, mostly because if one is a mixed shop one has to
    get *ultra-paranoid* (or at least, take the worst of the two
    for a working assumption) and therefore these are defined equal.
    Nevertheless, Linux tends to be more resistant to malware, which
    means that the corporation has to do the software equivalent of
    a honeypot muckout less often (if one compares desktop PCs to,
    say, portapotties -- and then remembers to wipe one's fingers
    afterwards :-) ).

[8] Development costs. If one has onsite engineering, one has to
    consider #2-#4 for the engineers as well, who have different
    requirements from the production system team. These include
    costs for development kits, time to development given
    those kits, and various other issues.

[9] Verification/QA costs. These might be folded into #7, or not.
    But we engineers aren't perfect. :-)

[10] Conversion costs. Under certain conditions one might
     take the one-time hit hinted at by #1, and record the
     actual outlay here. This presumably would include such
     things as:
     - hiring a clerk to re-enter everything
     - development of an automated tool that reads in Word and
       spits out RTF on the Microsoft system (this might
       be doable by something in VBScript/JScript, Windows
       and Word willing)
     - development/deployment of a tool that reads in Word and
       spits out something useful on the *Linux* system, something
       using or akin to the aforementioned POI, or maybe just
       something that automates OpenOffice.
     - hiring a fixup clerk to deal with the glitchies after a
       conversion effort
     - hiring a review clerk to ensure everything's correct

     (Note that "hiring" is not set in stone here;
     it's the putting on of a hat -- red or otherwise --
     by whoever's in charge of and/or doing the effort.
     All three might very well be handled by a competent
     software engineer, who types in what he needs to,
     automates what he can, and reviews the results.
     Note also that some of those documents may very well
     be on backup tapes.)

[11] Monitoring. Various issues ensue in a large production facility,
     among them the continued smooth operation of all of its parts.
     It's a bit like tweaking a small model airplane engine versus
     the complicated teardown/cleanout procedure for an aircraft
     engine used on a 767.

[12] Staff turnover costs. This slightly unusual entry is a must
     to factor in, though ideally this would be equal as well.
     However, Linux does have the "buzz" capability, and Microsoft
     has the "shine" put on by the work of thousands of unnamed
     marketeers, for just the right amount of gloss on its pamphlets,
     white papers, and other such literature ("litter"ature?).
     Which one's more interesting for a prospective employee,
     all other things being equal (which they probably aren't anyway)?
     Which one is more likely to help the employer keep the staff
     he already has?

[13] Research & estimation costs (project planning).

I say: don't be fooled by the hype on either side. :-)

There is a wildcard in all this: the traditional
Unix solutions such as Solaris, HP/UX, AIX, and Tru64.
(BSD/FreeBSD might be included as well, and FreeBSD is an
excellent freeware OS in its own right, along with its
distribution, AIUI. But this *is* a Linux advocacy group. :-) )

Some of these are one-time charges, some of these
recurring. For example, #3 is one-time (maybe), #2 is
either one time or recurring, depending on turnover.
#4 and #5 are definitely recurring. One hopes #10
is one-time, but already there's been a sea change from
HTML 2.0 (anyone else remember that?) to the current
XML/XSL/CSS/XHTML/RDF/etc. mixture.

There are also other solutions which I'm less familiar
with. VMS, however, I did use for many years (up to late
1989), and is still around. Mainframe operating systems
may also be around as well; the only ones I know of are
VM/CMS (from IBM), and MVS (a batch-processing system).

Tailor as required for one's own corporation. :-)

>
> Stan
>

[*] "compile" has two senses here, one of which is to
    gather and/or process information (e.g., to compile an
    FBI or other such dossier), the other to convert source
    code into object code using a tool (the compiler).
    It's an issue similar to "computer" which prior to
    the end of World War II or thereabouts was simply
    "one who computes", or "processor", which was "one
    who processes" (e.g., an insurance claims processor).

    One of the reasons distros are so attractive is that
    those supporting the distro have done most of the legwork
    for the individual using that distro.

-- 
#191, ewill3@earthlink.net
It's still legal to go .sigless.

• Previous message: George Jones IV: "Re: Sony Considers Linux for PS3 Hard Disk"
• In reply to: essteeaenn_at_worldbadminton.com: "Re: An Open Letter To The Linux Enthusiasts."
• Next in thread: Peter Köhlmann: "Re: An Open Letter To The Linux Enthusiasts."
• Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]