Mailing List Archive

Support open source code!


[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: tlug: PHT's HA Cluster License



>>>>> "ash" == Andrew S Howell <andy@example.com> writes:

>>>>> "Chris" == Christopher Sekiya <wileyc@example.com> writes:

    Chris> http://community.turbolinux.com/cluster/doc/faq/
    Chris> http://community.turbolinux.com/cluster/license/cluster-license.html

    Chris> $1000 per node?  Mix-and-match licenses?  Restricted by
    Chris> _license_ to kernel 2.0.36?

    ash> Interesting. I wonder how they get away with this?

I bet they don't.  Based on their desktop systems, I doubt their
configuration tools are all that robust.  That's obviously why they put
kernel restrictions etc in the license; control those variables!  Kill
the experimental animal if the results ain't right!  We'll see.

But my scheduled-to-be-replaced-by-Debian TL system can be overwritten
with no regrets, now.

Still, my thanks and kudos to Scott "who probably does not kiboze the
list archives" Stone, who busted his butt and did his best, even
though in his ignorance and youth he often disagreed with me.  :-)

    ash> I took a look at

    ash> http://proxy.iinchina.net/~wensong/ippfvs/

Interesting.

    ash> So, what are they licensing? Their monitoring tools? If they

Well, actually, it's the fault-tolerance code, implemented as a
separate daemon.  See the thread Chris references.  This would be a
big contribution, if and when they do.

But they're also presumably licensing service.  You saw what Craig
said, repeatedly: "should work".  The configuration and monitoring
tools are going to need babysitting for a while.  Note that PHT bumped
the usual convention for version numbering by 1; the prerelease eval
version is 1.x, the ready-for-prime-time-we-hope version is 2.  After
watching Scott work and knowing a little bit about how Craig works,
it's not likely to be a success by design (no comment intended on
either Craig or Scott, I'm talking about PHT's overall process,
cf. http://www.sei.cmu.edu/), assuming their skunkworks is organized
the way the desktop distribution is.  It's a great idea ->
code-and-fix project, I bet.  Notwithstanding claims of long interest
in the field.  (I'm sure the core daemon is well-designed and
implemented; I'm talking about the systems product.)

They're betting that by being first to market they can get a lot of
market share, and ride those revenues to improved tools, especially
configuration.  Building success on success.  My bet is that the
virtual server project will produce usable tools rather quickly.
Linux, Debian, and XFree86 have shown that you don't need a
proprietary angle to produce reasonably robust usable tools (although
dselect is getting way the f**k out of hand with over 3500 packages in
"potato"---but God save the Queen, it's still usable).  So PHT will
have to be ready to abandon their internally developed tools at any
moment, drop the proprietary license, and jump on the free software
bandwagon again (the tools don't have to be nearly as good as PHT's if
they are open source, I bet, so although PHT could have better tools
for a while, they may have trouble keeping a big enough gap).  Yo,
Cliff, Adrian, Rok, you guys listening?

[Since I wrote that, I read the thread Chris recommended.  Evidently
those tools will be open source from day one; they still should be
prepared to cut and run to supporting the free tools at any time---
maybe from day one as an alternative.  Yes, I know the internals are
different, I read the thread, ITUI,S.]

    ash> sold the tools seperately for a reasonable price, then I
    ash> would not mind. Restricting me to a particular kernel just
    ash> does not fly....

Yup.  And that audit provision is just plain insane.  For it to be
effective, they need to have root ("oh, I'm sorry, the partition where 
our subrosa server logs are kept is not mounted right now").  Idjit
lawyers.

And you guys can cut out the BS about MS-style monitoring tools.  That 
would have to be done by a cabal within PHT, I can't imagine that the
majority of developers there would condone it.  That could happen, of
course, but don't use it to beat anyone over the head with unless you
have reason to believe they know about it.

    ash> You can be sure I won't use it!

Now, now.  If either they're lucky and they get it right by accident,
or I'm wrong and they have a CMM Level 5 process going and they get it 
right by design, you should look at the business plusses and minuses.

And don't forget, they still produce a freely available Linux
distribution, with some of the best Asian support around.  True, they
can no longer claim to be a pure free software business; in fact those
Asian distributions may actually be loss-leaders.  But for the moment
we can't distinguish that from the oh-so-painful decision to milk a
cash cow so they can support the Linux user community better :-)

I think they're out to pay the current mortgage and add another 1000
sq ft to the current homestead (that would only take a couple thousand
licensed nodes, right?), myself, but I can't prove it ;-)

>>>>> "wile y" == Christopher Sekiya <wileyc@example.com> writes:

    wile y> This specific thread is at:  (omitted)
    wile y> ... and it certainly doesn't portray PHT in the best
    wile y> possible light.

No, it doesn't.  Pretty unfavorable, actually.

    wile y> (paraphrasing)

    wile y> PHT didn't credit the source of the kernel patches
    wile y> (visibly) until they were publically bashed for not doing
    wile y> so.

However, it's pretty clear from that thread that this division of PHT
does not have general development of free software at heart.  First
off, per-node licensing does not square with free software, although
per-node support contracts do.

The Aladdin license is an inappropriate comparison; the restrictions
in the Aladdin license are specifically directed at embedded systems,
where for practical purposes the end-user has none of the rights
specified in the GPL anyway, except that they can put a framed copy of
the source on the wall.  (That is, by definition embedded systems
manufacturers are violating the GPL.  But that's not RMS's
interpretation---it seems he sees the manufacturers as the end
users---so Peter Deutsch had to make a choice, and he chose to create
the Aladdin license.  This is inference from comments made by Peter, I
do not have an explicit statement from either Peter or RMS.)  This is
surely not true for HA clusters.

Throughout the honest (I see no reason to think they were malicious or
manipulating their policies to conceal a ruthless bent to profit)
errors that PHT made, we see a bias toward protecting their
proprietary advantage until that is consolidated, of course within
their interpretation of the spirit of free software.  They plan to
make it free, or partially free, anyway---but in the meantime,
"protect our proprietary advantage" through closed source and
marketing which treads the line of giving barely the credit that's due
is the watchword.  Not to mention the mention of a developer who did
not respond to queries regarding potential technical demerits of his
proposed patch.

The comparison with the thread in which the Lucid Emacs crowd battled
RMS is painful.  Huge egos all around, hugely arrogant power plays to
make the other side look bad, tempers out of control---but Lucid Emacs
was not only entirely GPL from day one (willingly so, I believe,
despite the huge improvements over both GNU Emacs 18.5x and Epoch),
but FSF-assigned to boot, as is today's XEmacs (except for some code
which is GPL but Sun Microsystems retains the copyright to).

-- 
University of Tsukuba                Tennodai 1-1-1 Tsukuba 305-8573 JAPAN
Institute of Policy and Planning Sciences       Tel/fax: +81 (298) 53-5091
__________________________________________________________________________
__________________________________________________________________________
What are those two straight lines for?  "Free software rules."
-------------------------------------------------------------------
Next Technical Meeting: June 19 (Sat), 18:30  place: Temple Univ.
*** Topic: Linux SMP/Quad Xeon Server
Next Nomikai: July 16 (Fri), 19:30 Tengu TokyoEkiMae 03-3275-3691
-------------------------------------------------------------------
more info: http://www.tlug.gr.jp        Sponsor: Global Online Japan


Home | Main Index | Thread Index

Home Page Mailing List Linux and Japan TLUG Members Links