How long does Sun have to live? (OT RANT)
Colin Bigam
question. What are they PLAYING at over there???
Today is the Big Quarterly Announcement. Woo. So far, no hint in the press
releases of any new workstations, except for some weird java stations which
are billed at $100/seat/year. Odd. StarOffice 6.1 finally has a release date
(Oct. 14), and a new name (StarOffice 7.0). WHY a new name, and why couldn't
they have bothered to give hints to anyone before this, about a release?
Personally, despite being in the pre-beta closed testing for SO6.1, I game
up waiting for a final release two months ago, and have been using openoffice
1.1.
Six months ago, when the Blade 1500 was looking like an imminent release,
there seemed to be some home for Sun on the workstation market. Now they've
been delayed so long (and still aren't released? We'll see about tomorrow),
that I don't think anyone really cares about them. Certainly not the
geophysical application vendors--Landmark, Geoquest, and all the others are
pushing hard for Linux workstations, and it makes sense. Most people who use
these programs have a ~$40k Sun workstation on their desk right beside their
$100 PC that's faster at most tasks anyways. If they could consolidate on one
machine, and spend an OUTRAGEOUS sum of (say) $8000 on it, then they'd be
saving over $30 000 per user on hardware costs, getting equivalent or better
performance, and also saving substantially on power costs.
Then there's storage. The low end stuff (3310 and 3350) are nice enough
boxes, but the new mid-range (6300 series) seems just as backwards as the T3.
None of them are as configurable as the venerable old RSM/Axx00 line, and
they're ultimately attached storage that needs a server. Even the Hitachi
is getting squeezed hard by Network Appliances and (to a lesser extent) EMC.
The OS? I love Solaris and (finally!!!) adding gnome to the default install
is great, but have you read their licensing requirements lately? They change
with every refresh it seems; and the latest requirement is that any CDs you
download and burn must have Sun approved labels attached to them, or you're
violating the licensing agreement.
Labelling burned CDs?? WTF were they smoking when they came up with THAT
idea?????
I'll leave their policy towards Linux alone, since I don't have a crazed
hermit to help me interpret what it is this afternoon. No doubt it will
have changed three times by tomorrow morning.
Oh yes, bugs! Has anyone at Sun noticed that people without paid support
contracts have no way of submitting bug reports? For that matter, there's
very little way of contacting Sun at all, if you're not paying support.
That means that those six developers who manage to qualify for the free
binary license program (or anyone else who conveniently ignores the license)
can't give Sun information that would help them make a better product!
Bill Joy leaving isn't going to help them either. The man may well be one
of the greatest visionaries alive today.
So after years of supporting Sun, running Solaris, owning Sparcs, and really
wanting to see this company succeed, I find myself asking WHAT ARE THEY
DOING OUT THERE??? Their hardware is slow and outrageously expensive, their
marketing seems to be a constantly-revising mess, Their OS is locked into
the craziest license I've ever seen, their apps are slow in coming and not
given much promotion. I know McNealy has been going on about N1, metered
computing, etc.; but I don't see how he intends to reform the company by
destroying it in the process.
Does anyone have any idea what the future of Sun is? There are too many
of their systems in mission critical places to fold entirely (at least
overnight), but I don't see how they can survive without being bought by
a company hostile to the Sparc architecture.
Or is Scott going to pull another rabbit out of his hat? He's done it
before, but I don't think I've ever seen things so grim for Sun before.
Colin
Fredrik Lundholm
Colin Bigam <spam...@go.com> wrote:
>OK, I know it's been beaten to death already, but I still have to ask the
>question. What are they PLAYING at over there???
>
>Today is the Big Quarterly Announcement. Woo. So far, no hint in the press
>releases of any new workstations, except for some weird java stations which
Woah, Sun released today the following interesting stuff (and more):
Sun Fire V440
Sun Fire V250
Sun Blade 1500
Sun Ray 1g+24" LCD
Sun Ray 1g
Announcing Sun Fire V880 1.2GHz (2003-09-16)
Control Station 2.0 (2003-09-16)
Sun One Webserver 6.1 (2003-09-16)
>Colin
Pretty interesting.
/wfr
Fredrik
--
Fredrik Lundholm
dol @ ce.chalmers.se
Chris Morgan
> Woah, Sun released today the following interesting stuff (and more):
>
[SNIP]
> Sun Blade 1500
[SNIP]
I see no sign of this. Can you quote your sources?
Chris
--
Chris Morgan
"Post posting of policy changes by the boss will result in
real rule revisions that are irreversible"
- anonymous correspondent
Kent Wormsdorf
>
> Woah, Sun released today the following interesting stuff (and more):
>
> Sun Fire V440
> Sun Fire V250
> Sun Blade 1500
> Sun Ray 1g+24" LCD
> Sun Ray 1g
> Announcing Sun Fire V880 1.2GHz (2003-09-16)
> Control Station 2.0 (2003-09-16)
> Sun One Webserver 6.1 (2003-09-16)
>
>>Colin
>
> Pretty interesting.
>
> /wfr
> Fredrik
Hi Fredrik
Do you know something I/We don't?!?!?!?!??
I can't finde anything.
Please enlighten me?
Greetings from a former Technical Presale from Denmark
Kent Wormsdorf
--
PS. (Sorry, only in danish.)
Er der nogen der mangler en mand med lidt viden om Sun HW og SW,
så må I gerne sende mig en E-mail.
Jeg er uddannet datamatiker og har senest arbejdet som teknisk presale konsulent.
Fredrik Lundholm
Chris Morgan <c...@mihalis.net> wrote:
>d...@ce.chalmers.se (Fredrik Lundholm) writes:
>
>> Woah, Sun released today the following interesting stuff (and more):
>>
>[SNIP]
>> Sun Blade 1500
>[SNIP]
>
>I see no sign of this. Can you quote your sources?
The INTRO's are posted to Suns reseller web and have been made "public".
According to Sun, the SB1500 can be ordered now and delivered this month.
Maybe we won't see it at store.sun.com until someones "reviels" it the
next following days though. In the US I expect it to have appeared in
the pricelists today and in Europe by tomorrow.
From the INTRO:
Launch Event/Analyst Briefing Date: September 16, 2003
Press Release Date: September 17, 2003
Revenue Release Date: September 09, 2003
General Availability Date: September 16, 2003
The revenue release date means that Sun is accepting orders since more
than a week.
Rich Teer
> > Sun Fire V440
> > Sun Fire V250
> > Sun Blade 1500
> > Sun Ray 1g+24" LCD
> > Sun Ray 1g
> > Announcing Sun Fire V880 1.2GHz (2003-09-16)
> > Control Station 2.0 (2003-09-16)
> > Sun One Webserver 6.1 (2003-09-16)
>
> Do you know something I/We don't?!?!?!?!??
>
> I can't finde anything.
>
> Please enlighten me?
No doubt, those products will be announced tomorrow.
Hmm, the Sun Blade 2500 doesn't seem to be on that list...
--
Rich Teer, SCNA, SCSA
President,
Rite Online Inc.
Voice: +1 (250) 979-1638
URL: http://www.rite-online.net
Alan Coopersmith
|Today is the Big Quarterly Announcement. Woo. So far, no hint in the press
|releases of any new workstations
Actually, it's this *week* is the big quarterly announcement - they've
spread them over three days at Sun Network. If you look at the
schedule, today was "Sun's Software Strategy" which means you wouldn't
expect hardware news. Wait until all three days are over before drawing
final conclusions.
(The schedule is at
http://sunnetwork.sun.com/sf2003/conf/schedule/index.en.jsp )
--
________________________________________________________________________
Alan Coopersmith al...@alum.calberkeley.org
http://www.CSUA.Berkeley.EDU/~alanc/ aka: Alan.Coo...@Sun.COM
Working for, but definitely not speaking for, Sun Microsystems, Inc.
Fredrik Lundholm
Rich Teer <rich...@rite-group.com> wrote:
>No doubt, those products will be announced tomorrow.
>Hmm, the Sun Blade 2500 doesn't seem to be on that list...
I was told by a Sun TSE that the reason for not releasing SB2500 now
was becuase of limited testing/validation resources.
Also that the test and validation team was severly overloaded.
(good thing no??)
Chris Morgan
> In article <86oexkh...@elrond.bloomberg.com>,
> Chris Morgan <c...@mihalis.net> wrote:
> >d...@ce.chalmers.se (Fredrik Lundholm) writes:
> >
> >> Woah, Sun released today the following interesting stuff (and more):
> >>
> >[SNIP]
> >> Sun Blade 1500
> >[SNIP]
> >
> >I see no sign of this. Can you quote your sources?
>
> The INTRO's are posted to Suns reseller web and have been made "public".
> According to Sun, the SB1500 can be ordered now and delivered this month.
> Maybe we won't see it at store.sun.com until someones "reviels" it the
> next following days though. In the US I expect it to have appeared in
> the pricelists today and in Europe by tomorrow.
Ok, I was only looking at the www.sun.com website. I don't know any
'reseller' info, i'm just some guy who looks after a Solaris
application.
Colin Bigam
> Actually, it's this *week* is the big quarterly announcement - they've
> spread them over three days at Sun Network. If you look at the
> schedule, today was "Sun's Software Strategy" which means you wouldn't
> expect hardware news. Wait until all three days are over before drawing
> final conclusions.
>
> (The schedule is at
> http://sunnetwork.sun.com/sf2003/conf/schedule/index.en.jsp )
OK, granted. Still, I don't see much in there that implies hardware other
than Greg Popadopolous' talk on "The New System." (which sounds to my jaded
ears like another way of saying, "it's a whole new paradigm" instead of
"here are the machines you've been waiting months for")
And besides--are these machines going to be particularly relevant? If we
assume Fredrik is correct in his list of new stuff coming out, then we
get things like the V880 getting 1.2GHz processors. Oh boy, I can hardly
wait to get these things which absolutely blow the doors off of Intel/AMD
processors at the same MHz. Pity they don't hold up to a 3+GHz x86 though.
Random aside: Just how many forms of the US-III processor ARE there? The
Blade 1k/2k/V280 has one, the V480/880 has another, the Fx800 have a third,
and now the V1280 makes four.
The Blade 1500 looks like it'll be a decent US-IIIi PC. Passable capacity,
passable speed, nothing fancy. If they sell these for about $2-3k including
a PGX64 framebuffer, they might stand a chance. Unfortunately, I don't
forsee prices like that.
Regardless of all of this hardware (mostly workstation) talk, is the fact
that Sun is feeling...bogged down, for lack of a better term. They're buying
hardware because they can't design their own quickly enough, and then killing
it off because it's questionable technology, or at least not something that
works with Sun's model.
It's more than hardware. It feels like Sun has lost any vision they might
have had, or possibly that they're so blinded by some new vision, that
they've forgotten how to run a business. I feel like my favourite Unix
company is sick, and I'm starting to believe that it might be terminal.
Can Sun pull it all together?
Colin
Rich Teer
> I was told by a Sun TSE that the reason for not releasing SB2500 now
> was becuase of limited testing/validation resources.
> Also that the test and validation team was severly overloaded.
> (good thing no??)
I keep saying to Sun "Send me test hardware, I'll do it for free!".
But they don't send me the toys^Wtest hardware. :-(
Note to Alan Coopersmith: I promist it would help me write faster... :-)
Rich Teer
> The Blade 1500 looks like it'll be a decent US-IIIi PC. Passable capacity,
> passable speed, nothing fancy. If they sell these for about $2-3k including
> a PGX64 framebuffer, they might stand a chance. Unfortunately, I don't
> forsee prices like that.
The SB 1500 starts at $2995 list, an would imagine that that
includes the XVR-100 frame buffer (the PGX-64's replacement).
Dunno what else it includes, though, as the Sun Store link was
broken when I tried it.
And Mark Tolliver's "Live" webcast still hasn't started. :-(
> they've forgotten how to run a business. I feel like my favourite Unix
> company is sick, and I'm starting to believe that it might be terminal.
I echo some of your sentiment...
> Can Sun pull it all together?
They've dug themselves out of it in the past, and I'm hopeful
they'll be able to pull they same trick off again.
klas_sundberg
Originally posted by Rich Teer
> On 17 Sep 2003, Colin Bigam wrote:
>
>
> Dunno what else it includes, though, as the Sun Store link was
> broken when I tried it.
>
From the SUN home page:
Sun Blade 1500 Workstation
Using the new 1-GHz UltraSPARC IIIi processor, the price/performance-
leading Sun Blade 1500 workstation offers high-performance 64-bit
computing ideal for a wide range of business and technical environments.
The Sun Blade 1500 workstation is equipped with a robust set of
workstation features, including support for 4 GB of DDR memory, up to
two internal 80-GB Ultra ATA100 hard drives, five PCI slots, and support
for high-performance graphics such as the Sun XVR-100 or Sun XVR-500
graphics accelerators. The Sun Blade 1500 workstation also offers built-
in USB 1.1, USB 2.0 and IEEE 1394a FireWire ports and on-board Gigabit
Ethernet (10/100/1000Base-T).
SMALL
UltraSPARC IIIi Processor 1 @ 1 GHz
On-Chip L2 Cache 1 MB
Graphics Accelerator Sun XVR-100
Memory 512 MB (2 @ 256-MB DIMMS)
7200 RPM IDE Disk Drive 1 @ 80 GB
Ethernet Port 1 @ 10/100/1000 BaseT
USB Port 7
IEEE 1394 Port 2
Parallel Port 1
Serial Port 2
PCI Slot 5
16x DVD-ROM Drive 1
Smart Card Reader 1
Operating System Solaris 8 HW 5/03 Pre-Installed
List Price $2,995.00
LARGE
UltraSPARC IIIi Processor 1 @ 1 GHz
On-Chip L2 Cache 1 MB
Graphics Accelerator Sun XVR-500
Memory 1 GB (2 @ 512-MB DIMMS)
7200 RPM IDE Disk Drive 1 @ 80 GB
Ethernet Port 1 @ 10/100/1000 BaseT
USB Port 7
IEEE 1394 Port 2
Parallel Port 1
Serial Port 2
PCI Slot 5
16x DVD-ROM Drive 1
Smart Card Reader 1
Operating System Solaris 8 HW 5/03 Pre-Installed
Call Sun
List Price $3,995.00
--
Posted via http://dbforums.com
Anthony Mandic
> Alan Coopersmith <al...@alum.calberkeley.org> wrote:
> >
> > http://sunnetwork.sun.com/sf2003/conf/schedule/index.en.jsp
> OK, granted. Still, I don't see much in there that implies hardware other
> than Greg Popadopolous' talk on "The New System." (which sounds to my jaded
> ears like another way of saying, "it's a whole new paradigm" instead of
> "here are the machines you've been waiting months for")
Oh, great. The new acronym for Project Orion is "JES."
-am © 2003
Anthony Mandic
>
> Sun Blade 1500 Workstation
>
> Using the new 1-GHz UltraSPARC IIIi processor, the price/performance-
> leading Sun Blade 1500 workstation offers high-performance 64-bit
> computing ideal for a wide range of business and technical environments.
>
> The Sun Blade 1500 workstation is equipped with a robust set of
> workstation features, including support for 4 GB of DDR memory, up to
What kind of bullcrap is that? My new Apple G5 (on order) will
have two internal 250 Gb serial ATA drives on independent buses
and be able to max out at 8 Gb memory...
-am © 2003
klas_sundberg
Originally posted by Anthony Mandic
> What kind of bullcrap is that? My new Apple G5 (on order) will
> have two internal 250 Gb serial ATA drives on independent buses
> and be able to max out at 8 Gb memory...
>
> -am © 2003
Well to be fair
http://www.sun.com/desktop/sunblade1500/details.html]8 Gb will be
supported in Q1 2004[/url].
[rant]
But other then that, I do have to agree that the specifications could be
made beefier without causing cannibalisation of the higher spec (higher
profit margin) SB 2000. Such as a S-ATA controller, an UPA-slot, a PCI-X
controller instead of the 64-bit PCI one. And why not integrated
bluetooth
and WiFi so that one could with ease use cordless keyboards and mice and
move medium sized files to one's laptop?
Not to mention the questionable visual ergonomics of it,
that http://www.sun.com/desktop/sunblade1500/images/I1_hw_sunblade-
_1500_lg.jpgbig red blob on the middle of the case is as
ergonomically bad as
bad can be IMO, not to mention how much prettier it would be without it,
with a more clean and visually coherent design.
[/rant]
Steve
Non-ECC memory at that. Personally, I gag at anything called
a "workstation" with IDE disks!
Apple isn't perfect either. Note I just purchased a 1.33Ghz 17"
PowerBook this week - it is really nice. But I think it will
primarily be for what I purchased it for - to edit movies and
make DVDs. It isn't a solution for everything.
I imagine the IIIi delays mean that the motherboard was designed
long before it was announced. Now that IIIi are available, I
would expect the next generation to be within a year - where
there will again be an opportunity to apply the latest toys.
Just another lurker
> Oh boy, I can hardly
> wait to get these things which absolutely blow the doors off of Intel/AMD
> processors at the same MHz. Pity they don't hold up to a 3+GHz x86 though.
What's all this about MHz? Nobody who buys Sun products really cares
much about how it compares with an Intel box speedwise. There are many
other qualities to the SPARC processor family (SMP for example).
> Random aside: Just how many forms of the US-III processor ARE there? The
> Blade 1k/2k/V280 has one, the V480/880 has another, the Fx800 have a third,
> and now the V1280 makes four.
I only count three: US III (replaced by the Cu version now), US III Cu,
and US IIIi. What's the fourth one?
> The Blade 1500 looks like it'll be a decent US-IIIi PC. Passable capacity,
> passable speed, nothing fancy. If they sell these for about $2-3k including
> a PGX64 framebuffer, they might stand a chance. Unfortunately, I don't
> forsee prices like that.
Hmmm... the Blade 1500 is already available through Sun's online store.
The price for one box with 1GHz proc, 512MB memory, an 80GB drive, and
a Sun XVR-100 framebuffer sets you back by $3,000. Not so shabby.
/d
Frank Cusack
> Hmmm... the Blade 1500 is already available through Sun's online
> store. The price for one box with 1GHz proc, 512MB memory, an 80GB
> drive, and a Sun XVR-100 framebuffer sets you back by $3,000. Not so
> shabby.
Seems shabby to me, when I can get a rougly equivalent HP xw4100 for $1k.
Please correct me if the xw4100 isn't roughly equivalent.
/fc
Chris Morgan
(donning nomex suit)
Actually it's not even roughly equivalent, as it would probably
perform quite a bit better on some tasks :)
On the other hand, for me the Sun Blade 1500 can run SPARC/Solaris,
which no HP can at any price. For those parts of my job that involve
sparc/solaris, that is clearly a huge advantage!
Colin Bigam
Well, since I started this, I figure I can defend Sun a bit here. :-)
(snip)
> But other then that, I do have to agree that the specifications could be
> made beefier without causing cannibalisation of the higher spec (higher
> profit margin) SB 2000. Such as a S-ATA controller, an UPA-slot, a PCI-X
> controller instead of the 64-bit PCI one. And why not integrated
> bluetooth
> and WiFi so that one could with ease use cordless keyboards and mice and
> move medium sized files to one's laptop?
Are Serial ATA and PCI-X compatable with older hardware? If not, I can
see Sun's point of view--they don't want to be making their own
middle-of-the-road workstations unable to use the parts the've already
got. (Not so much a concern for the drives, but definitely for the PCI cards).
The lack of a UPA slot, and the presence of only a single 64b/64MHz PCI
slot suggests to me that this will never be a workstation though. If it
could accept two XVR-1200 cards (and run them at full spec), then I'd
expect this to take over a chunk of the Blade2k market by storm. Without
that ability (and in fact with the XVR-500 being the maximum card supported
ATM), the 1500 is a desktop machine. Not a bad one mind you, and the price
is right.
> Not to mention the questionable visual ergonomics of it,
Well, um...
Yeah. Possibly even worse than the Ultra10.
Colin
Chris Morgan
> The lack of a UPA slot, and the presence of only a single 64b/64MHz PCI
> slot suggests to me that this will never be a workstation though. If it
> could accept two XVR-1200 cards (and run them at full spec), then I'd
> expect this to take over a chunk of the Blade2k market by storm. Without
> that ability (and in fact with the XVR-500 being the maximum card supported
> ATM), the 1500 is a desktop machine. Not a bad one mind you, and the price
> is right.
Sun seems to be moving away from UPA. The Blade 150's cpu doesn't even
support it, even though it's labelled a US-IIi, and older US-IIis did.
I think US-IIIi supports UPA on-chip, but none of the systems it comes
in have it.
Derek Snider
> OK, I know it's been beaten to death already, but I still have to ask the
> question. What are they PLAYING at over there???
> I'll leave their policy towards Linux alone, since I don't have a crazed
> hermit to help me interpret what it is this afternoon. No doubt it will
> have changed three times by tomorrow morning.
> Does anyone have any idea what the future of Sun is? There are too many
> of their systems in mission critical places to fold entirely (at least
> overnight), but I don't see how they can survive without being bought by
> a company hostile to the Sparc architecture.
If Sun were wise, they would embrace Linux as IBM has done.
This doesn't mean they have to get rid of Solaris -- they can release
Solaris 10L to compete with IBM's AIX 5L (the L being for Linux
compatability).
They would also work hard to ensure that Linux is able to take full
advantage of their hardware, and sell ALL their mainframes with a
choice of Linux as the OS (not just on the x86 based hardware).
Sun would sell more UltraSparcs if users had their choice of Solaris
or Linux -- even if 100% of the customers wanted Solaris, they'd get
more customers because the CHOICE was available.
Linux may be responsible for putting a huge dent into Sun's server
business, but Linux is going to end up being the only thing that will
save them.
Rich Teer
> They would also work hard to ensure that Linux is able to take full
> advantage of their hardware, and sell ALL their mainframes with a
> choice of Linux as the OS (not just on the x86 based hardware).
Who on EARTH would be stupid enough to run Linux on a
$10 million SF 15K?
> Sun would sell more UltraSparcs if users had their choice of Solaris
> or Linux -- even if 100% of the customers wanted Solaris, they'd get
> more customers because the CHOICE was available.
Give me a break. Many people buy SPARC machines BECAUSE of
Solaris. Aside from hobbyist hack value, Linux on SPARC is,
IMHO, a waste of time.
> Linux may be responsible for putting a huge dent into Sun's server
> business, but Linux is going to end up being the only thing that will
> save them.
Keep dreaming, troll.
Frank Cusack
> If Sun were wise, they would embrace Linux as IBM has done.
Nooo!!
Linux is death-come-knocking for Sun. Rather than embrace Linux, they
should make Solaris as hacker friendly as Linux is. THEN we'll see
who wins the OS wars.
Sun needs to support x86 a lot better (only because price/performance is
so much better), but that does NOT mean Linux. Or, they need to bring
SPARC price/performance into line with x86, at the low/middle end.
Like Apple, Sun is a hardware AND software company. Unlike Apple, they
don't have attractive hardware from the price/performance perspective.
Since I doubt Sun can bring the low end SPARC to compete with x86
(demonstrated by this week's hardware announcements), I'd say that
the software arm needs to ignore the hardware arm's SPARC agenda and
treat x86 as a first class platform.
And at the top end, where they shine, Sun has a real problem in that
clusters are here. The large single system image is becoming more of
a niche every quarter. How do you build the best clusters? With
hardware that gives the best price/performance. x86 now, probably
will become opteron soon.
Smart people today only run on SPARC if they *have to*. Sun needs to
fix that.
Companies like IBM and HP can afford to embrace Linux because they
have a large x86 hardware business. They sell Windows bundled with
their hardware, for chrissake.
> This doesn't mean they have to get rid of Solaris -- they can release
> Solaris 10L to compete with IBM's AIX 5L (the L being for Linux
> compatability).
They already have Linux compatibility -- lxrun.
> Sun would sell more UltraSparcs if users had their choice of Solaris
> or Linux -- even if 100% of the customers wanted Solaris, they'd get
> more customers because the CHOICE was available.
Not at all. Folks that want Linux are NOT going to pay the substantial
price penalty to run on SPARC hardware. However, the reverse is not
true -- folks that want to run Solaris are willing to pay the reliability
or scaling penalty to run on cheap x86 hardware ... just buy two and
load balance.
/fc
klas_sundberg
Originally posted by Colin Bigam
>
> Are Serial ATA and PCI-X compatable with older hardware?
> If not, I can see Sun's point of view
PCI-X is backwards compatible with PCI 64 bit cards and
(like the PCI 64 bit) newer specification PCI 32bit ones.
Having S-ATA and a regular ATA "functions" on the same
bus/chip should not pose any problems,
I think the connectors should have to be different thought.
>
> > Not to mention the questionable visual ergonomics of it,
>
> Well, um...
>
> Yeah. Possibly even worse than the Ultra10.
>
> Colin
That was kind of a stuck up way of calling it fugly wasn't it :)
But looks and flaws aside, if SUN lowers the price by 500$
then this defenetly is one nice SUN box.
Barton Fisk
> spam...@go.com (Colin Bigam) wrote in message
> news:<73c76674.03091...@posting.google.com>...
>> OK, I know it's been beaten to death already, but I still have to ask
>> the
>> question. What are they PLAYING at over there???
>
>> I'll leave their policy towards Linux alone, since I don't have a crazed
>> hermit to help me interpret what it is this afternoon. No doubt it will
>> have changed three times by tomorrow morning.
This is the major problem. There have been so many erratic changes in
software policies and strategies in general, that it leaves the customer
uncertain. There is a perception that the only sure bet with Sun is Solaris
on Sparc.
>
>> Does anyone have any idea what the future of Sun is? There are too many
>> of their systems in mission critical places to fold entirely (at least
>> overnight), but I don't see how they can survive without being bought by
>> a company hostile to the Sparc architecture.
Stranger things have happened. We could speculate.
>
> If Sun were wise, they would embrace Linux as IBM has done.
>
Oh, you mean the markitecture they show on their commercials? Great
marketing.
> They would also work hard to ensure that Linux is able to take full
> advantage of their hardware, and sell ALL their mainframes with a
> choice of Linux as the OS (not just on the x86 based hardware).
Why? So you can load the latest update to your 15k firmware from an rpm? I
cringe at the thought. Seriously this makes no sense.
> Sun would sell more UltraSparcs if users had their choice of Solaris
> or Linux -- even if 100% of the customers wanted Solaris, they'd get
> more customers because the CHOICE was available.
More marketing? Applications drive the platform.
>
> Linux may be responsible for putting a huge dent into Sun's server
> business, but Linux is going to end up being the only thing that will
> save them.
>
You had better watch it, SCO is botting these groups, looking for companies
to invoice for license fees. :-)
--
Barton Fisk SCNA SCSA
Remove .NOSPAM to reply via emai
---------
Using M2, Opera's revolutionary e-mail client: http://www.opera.com/m2/
Beardy
> Ok, I was only looking at the www.sun.com website. I don't know any
> 'reseller' info, i'm just some guy who looks after a Solaris
> application.
>
> Chris
Chris, we are all people who in some way look after "the" Solaris
application. Don't try to defend yourself - bite them before they bite
you! Chances are you'll be right...
Fr3aK3rMaN
> On 18 Sep 2003 14:29:28 -0700 de...@idirect.com (Derek Snider) wrote:
>
>>If Sun were wise, they would embrace Linux as IBM has done.
>
>
> Nooo!!
>
> Linux is death-come-knocking for Sun. Rather than embrace Linux, they
> should make Solaris as hacker friendly as Linux is. THEN we'll see
> who wins the OS wars.
>
> Sun needs to support x86 a lot better (only because price/performance is
> so much better), but that does NOT mean Linux. Or, they need to bring
> SPARC price/performance into line with x86, at the low/middle end.
>
> Like Apple, Sun is a hardware AND software company. Unlike Apple, they
> don't have attractive hardware from the price/performance perspective.
> Since I doubt Sun can bring the low end SPARC to compete with x86
> (demonstrated by this week's hardware announcements), I'd say that
> the software arm needs to ignore the hardware arm's SPARC agenda and
> treat x86 as a first class platform.
>
> And at the top end, where they shine, Sun has a real problem in that
> clusters are here. The large single system image is becoming more of
> a niche every quarter. How do you build the best clusters? With
> hardware that gives the best price/performance. x86 now, probably
> will become opteron soon.
>
> Smart people today only run on SPARC if they *have to*. Sun needs to
> fix that.
>
I agree with you on that. I run SPARC at home because it is so much
better in quite some ways.... but a lot of people can't afford it.
> Companies like IBM and HP can afford to embrace Linux because they
> have a large x86 hardware business. They sell Windows bundled with
> their hardware, for chrissake.
>
>
>>This doesn't mean they have to get rid of Solaris -- they can release
>>Solaris 10L to compete with IBM's AIX 5L (the L being for Linux
>>compatability).
>
>
> They already have Linux compatibility -- lxrun.
>
>
>>Sun would sell more UltraSparcs if users had their choice of Solaris
>>or Linux -- even if 100% of the customers wanted Solaris, they'd get
>>more customers because the CHOICE was available.
>
>
> Not at all. Folks that want Linux are NOT going to pay the substantial
> price penalty to run on SPARC hardware. However, the reverse is not
> true -- folks that want to run Solaris are willing to pay the reliability
> or scaling penalty to run on cheap x86 hardware ... just buy two and
> load balance.
>
> /fc
That's indeed... i want to run Solaris at any cost so i buy SPARC!!
Phillip Fayers
>On 18 Sep 2003 14:29:28 -0700 de...@idirect.com (Derek Snider) wrote:
>> If Sun were wise, they would embrace Linux as IBM has done.
>Nooo!!
>Linux is death-come-knocking for Sun. Rather than embrace Linux, they
>should make Solaris as hacker friendly as Linux is. THEN we'll see
>who wins the OS wars.
>Sun needs to support x86 a lot better (only because price/performance is
>so much better), but that does NOT mean Linux. Or, they need to bring
>SPARC price/performance into line with x86, at the low/middle end.
Sun do seem to have (finally) got the idea with Solaris on x86.
At their recent event they started to make noise about how Solaris on
x86 makes better use of things like the Xeon/Pentium Hyperthreading than
Linux does. I'm disappointed that the "Mad Hatter" desktop isn't
available running on Solaris as well as Linux, maybe we can look forward
to that in the future?
>Like Apple, Sun is a hardware AND software company. Unlike Apple, they
>don't have attractive hardware from the price/performance perspective.
>Since I doubt Sun can bring the low end SPARC to compete with x86
>(demonstrated by this week's hardware announcements), I'd say that
>the software arm needs to ignore the hardware arm's SPARC agenda and
>treat x86 as a first class platform.
There are rumours that Sun will start to use AMD Opterons in systems,
possibly in workstations. They've also talked about porting Solaris to
Opteron.
...
>Smart people today only run on SPARC if they *have to*. Sun needs to
>fix that.
Definately. They've given up on getting high performance from a single
CPU thread, which will eventually (shortly?) mean the death of UltraSPARC
as a workstation CPU. Look at Suns own systems. The Xeon based 1U and 2U
systems are faster and cheaper that the UltraSPARC IIIi based alternatives.
--
Phillip Fayers School of Physics & Astronomy, Cardiff University (*).
[ (*) - the official trading name of the University of Wales, Cardiff. ]
P.Fa...@astro.cf.ac.uk http://www.astro.cf.ac.uk/pub/Phillip.Fayers/
Tel: +44 (0)29 2087 5282 Attribute these comments to me, not UWC.
Scott Howard
> Not to mention the questionable visual ergonomics of it,
>
> that http://www.sun.com/desktop/sunblade1500/images/I1_hw_sunblade-
> _1500_lg.jpgbig red blob on the middle of the case is as
> ergonomically bad as
It's actually not as bad as it looks. I did do a double-take when I first
saw one (before I'd seen a photo), but "in the flesh" they do actually look
quite good.
Scott
Scott Howard
>> The Sun Blade 1500 workstation is equipped with a robust set of
>> workstation features, including support for 4 GB of DDR memory, up to
>> two internal 80-GB Ultra ATA100 hard drives...
>
> What kind of bullcrap is that? My new Apple G5 (on order) will
> have two internal 250 Gb serial ATA drives on independent buses
> and be able to max out at 8 Gb memory...
And you'll just have to cross your fingers and hope you never have a
bit-flipped in that non-parity, non-ECC protected memory on the G5. If you
o (and experience shows that you almost certainly will at some point
in time) then odds are you've just got yourself some of that nice stuff
called "silent data corruption".
Scott.
Scott Howard
> If Sun were wise, they would embrace Linux as IBM has done.
So, out of interest how well does Linux run on a 200 CPU machine?
Scott
Mr. Johan Andersson
On Fri, 19 Sep 2003, Phillip Fayers wrote:
>
> Sun do seem to have (finally) got the idea with Solaris on x86.
> At their recent event they started to make noise about how Solaris on
> x86 makes better use of things like the Xeon/Pentium Hyperthreading than
> Linux does. I'm disappointed that the "Mad Hatter" desktop isn't
> available running on Solaris as well as Linux, maybe we can look forward
> to that in the future?
I believe you will.
> There are rumours that Sun will start to use AMD Opterons in systems,
> possibly in workstations. They've also talked about porting Solaris to
> Opteron.
Havent heard that yet...
I sure like the prices that are coming in on the servers though...
f.ex. the V440
As I see, it comes in three standard configurations.
2 x 1.062GHz/1Mb Cache, 4Gb Ram (8x512Mb), 4 36Gb 10kRPM SCSI, DVD and
dual power at ~$9k
4 x 1.062GHz/1Mb Cache, 8Gb Ram (16x512Mb), 4x36Gb 10kRPM SCSi, DVD and
dual power at ~$14k
and
4 x 1.28GHz/1Mb Cache, 16Gb Ram (16x1Gb), 4x36Gb 10kRPM SCSI, DVD and dual
power at ~$23k
My prices are filtered through the swedish ones, so they might vary in
other parts of the world, but a comparation we made to other vendors with
the same hw configs theese systems were at half the price.
Yes I know is not a Workstation, but the MadHatter concept (Java
Enterprise Station?) will work for the general WS users, actually quite nice
even though I personally use Mandrake normaly I sure like this one.
As I understand it it will come on the SPARC platform too, but there I
think it will be solaris/madhatter, not linux, to some peoples joy and
others sorrow.
/Johan A
Anthony Mandic
I've seen that happen on a Sun coming off of a Sun-approved
disk drive, where the controller was silently screwing up
the data.
We were able to track it down because it involved an admin
making an exact duplication of a quiescent disk, moving it
to another system, getting failures.
There is no magic bullet.
Well, maybe good backups.
-am © 2003
Rich Teer
> I agree with you on that. I run SPARC at home because it is so much
> better in quite some ways.... but a lot of people can't afford it.
Looking at the price of Ultra 1s and ULtra 30s on Ebay, I don't think
cost is a reasonable excuse not to use SPARC hardware. OK, they're
not state of the art, but for many (Home based) tasks, they're "good
enough".
Rich Teer
> And you'll just have to cross your fingers and hope you never have a
> bit-flipped in that non-parity, non-ECC protected memory on the G5. If you
> o (and experience shows that you almost certainly will at some point
> in time) then odds are you've just got yourself some of that nice stuff
> called "silent data corruption".
Hey, maybe the G5 uses the Broadcom GBE chip, and the designers
were hoping that the non-ECC memory would undo the silent corruption
caused by the iffy GBE. :-)
Paul Eggert
It depends on what you want to run. But if you're talking the
SPECint_rate_base2000 benchmark, the best published GNU/Linux score is
601 (the SGI Altix, with 64 1.3 GHz Itanium 2 processors).
In contrast, the best published Sun score is only
430 (Sun Fire 15K, with 72 1.2 GHz UltraSPARC III cu processors).
Fujitsu-Siemens beats Sun, but their best score is only
540 (Fujitsu-Siemens PRIMEPOWER2000, with 128 675 MHz SPARC64 GP processors).
So, at least by this benchmark, both Sun and Solaris are behind GNU/Linux
in terms of running on large multiple-CPU boxes. Obviously your mileage
will vary, and you need to test using your own application.
By the way, the current champion score at this particular benchmark is
1344 (SGI Origin 3800, with 256 600 MHz MIPS R14000A processors,
running IRIX). A 64-processor HP-UX Superdome server running HP-UX
also beats the best GNU/Linux score, by quite a bit (it scored 904).
Derek Snider
> On 18 Sep 2003, Derek Snider wrote:
> > They would also work hard to ensure that Linux is able to take full
> > advantage of their hardware, and sell ALL their mainframes with a
> > choice of Linux as the OS (not just on the x86 based hardware).
> Who on EARTH would be stupid enough to run Linux on a
> $10 million SF 15K?
This clearly shows how behind the times you guys are.
The fastest Linux-based supercomputer ranks #3 on the top 500
supercomputer list. The fastest Solaris-based supercomputer ranks
#211 running on your precious "SF 15K". (http://www.top500.org/)
Here's some articles that may be of interest to you...
IBM Linux Supercomputer To Power Life Sciences Grid:
http://www.gridcomputingplanet.com/news/article.php/2244781
Linux supercomputer rocks down under:
http://www.vnunet.com/News/1142962
Linux Supercomputer Howls:
http://www.wired.com/news/technology/0,1282,35113,00.html
Derek Snider
> On 18 Sep 2003 14:29:28 -0700 de...@idirect.com (Derek Snider) wrote:
> > If Sun were wise, they would embrace Linux as IBM has done.
> Nooo!!
Yes.
> Linux is death-come-knocking for Sun. Rather than embrace Linux, they
> should make Solaris as hacker friendly as Linux is. THEN we'll see
> who wins the OS wars.
There's no way that Solaris would ever be able to compete on that
issue.
> Sun needs to support x86 a lot better (only because price/performance is
> so much better), but that does NOT mean Linux. Or, they need to bring
> SPARC price/performance into line with x86, at the low/middle end.
They need to do exactly what I said -- which is exactly what IBM has
done.
> Like Apple, Sun is a hardware AND software company. Unlike Apple, they
> don't have attractive hardware from the price/performance perspective.
> Since I doubt Sun can bring the low end SPARC to compete with x86
> (demonstrated by this week's hardware announcements), I'd say that
> the software arm needs to ignore the hardware arm's SPARC agenda and
> treat x86 as a first class platform.
IBM is also a hardware AND software company... but they're finally
wising up to the concept that the days of OS software being a
mainstream cash product are coming to a close. Within the next ten
years buying a proprietary closed-source OS will be nothing buy an
unpleasant memory.
> And at the top end, where they shine, Sun has a real problem in that
> clusters are here. The large single system image is becoming more of
> a niche every quarter. How do you build the best clusters? With
> hardware that gives the best price/performance. x86 now, probably
> will become opteron soon.
Much sooner than you think.
> > This doesn't mean they have to get rid of Solaris -- they can release
> > Solaris 10L to compete with IBM's AIX 5L (the L being for Linux
> > compatability).
> They already have Linux compatibility -- lxrun.
IBM's AIX 5L allows for source-level Linux compatability whereby
making cross-platform compatable code a breeze. Code designed to be
built under Linux will compile under AIX 5L without a hitch.
> Folks that want Linux are NOT going to pay the substantial price penalty
> to run on SPARC hardware.
You're a little confused here. People don't want to run Linux because
it's cheap... they want to run it because it's free -- and I don't
mean the price... I mean that it's GPL code. Free as in freedom.
Confused? This link will explain:
http://www.gnu.org/philosophy/free-sw.html
Steve
wrote:
> Frank Cusack <fcu...@fcusack.com> wrote in message news:...
...
>> Like Apple, Sun is a hardware AND software company. Unlike Apple, they
>> don't have attractive hardware from the price/performance perspective.
>> Since I doubt Sun can bring the low end SPARC to compete with x86
>> (demonstrated by this week's hardware announcements), I'd say that
>> the software arm needs to ignore the hardware arm's SPARC agenda and
>> treat x86 as a first class platform.
>
> IBM is also a hardware AND software company... but they're finally
> wising up to the concept that the days of OS software being a
> mainstream cash product are coming to a close. Within the next ten
> years buying a proprietary closed-source OS will be nothing buy an
> unpleasant memory.
Keep smoking whatever you are smoking! It must be good shit.
Sun is getting back on the right track - people want cost-
effective *SYSTEMS*. Not piston rings and spark plugs.
IBM is pathetic at a broad spectrum of fully compatible
*SYSTEMS*.
Utility computing is coming - and Sun has a much better
basis for a good story.
Alan Coopersmith
|IBM's AIX 5L allows for source-level Linux compatability whereby
|making cross-platform compatable code a breeze. Code designed to be
|built under Linux will compile under AIX 5L without a hitch.
That's truly amazing since much Linux code won't even build under
different versions/distros of Linux.
Nevertheless, Solaris contains many of the same API's already, and most
popular open source code is already fully portable between Linux and
Solaris.
--
________________________________________________________________________
Alan Coopersmith al...@alum.calberkeley.org
http://www.CSUA.Berkeley.EDU/~alanc/ aka: Alan.Coo...@Sun.COM
Working for, but definitely not speaking for, Sun Microsystems, Inc.
Rich Teer
> There's no way that Solaris would ever be able to compete on that
> issue.
I agree - Sun would be foolish to opnely publish the Solaris
source code.
> They need to do exactly what I said -- which is exactly what IBM has
> done.
Yeah, but the difference is that AIX is a POS, unlike Solaris.
> IBM's AIX 5L allows for source-level Linux compatability whereby
> making cross-platform compatable code a breeze. Code designed to be
> built under Linux will compile under AIX 5L without a hitch.
Whoopie! Code that is written standards on Solaris will compile
without a hitch on Linux too. What's your point?
> You're a little confused here. People don't want to run Linux because
I think you're wrong here. Hobbyists aside, people (i.e.,
businesses) use Linux because it is percieved as being cheap.
> it's cheap... they want to run it because it's free -- and I don't
> mean the price... I mean that it's GPL code. Free as in freedom.
If that's the case, why hasn't *BSD taken the world by storm?
It's as free as the Linux kernel, and (most of) the userspace
apps are the SAME.
Paul Floyd
[snip]
>> This doesn't mean they have to get rid of Solaris -- they can release
>> Solaris 10L to compete with IBM's AIX 5L (the L being for Linux
>> compatability).
>
> They already have Linux compatibility -- lxrun.
lxrun isn't great. I wouldn't want to rely on it for anything important.
It certainly doesn't get updated to keep up the pace with Linux
developments.
A bientot
Paul
--
Paul Floyd http://paulf.free.fr (for what it's worth)
Surgery: ennobled Gerald.
Joerg Schilling
Paul Floyd <ro...@127.0.0.1> wrote:
>>> This doesn't mean they have to get rid of Solaris -- they can release
>>> Solaris 10L to compete with IBM's AIX 5L (the L being for Linux
>>> compatability).
>>
>> They already have Linux compatibility -- lxrun.
>
>lxrun isn't great. I wouldn't want to rely on it for anything important.
>It certainly doesn't get updated to keep up the pace with Linux
>developments.
This is why I like to see 'lkp' from SCO Unixware on Solaris.
--
EMail:jo...@schily.isdn.cs.tu-berlin.de (home) Jörg Schilling D-13353 Berlin
j...@cs.tu-berlin.de (uni) If you don't have iso-8859-1
schi...@fokus.fraunhofer.de (work) chars I am J"org Schilling
URL: http://www.fokus.fraunhofer.de/usr/schilling ftp://ftp.berlios.de/pub/schily
Dragan Cvetkovic
> On 19 Sep 2003, Derek Snider wrote:
>
>
>> They need to do exactly what I said -- which is exactly what IBM has
>> done.
>
> Yeah, but the difference is that AIX is a POS, unlike Solaris.
POS probably not meaning "point of sale" here :-)
Bye, Dragan
--
Dragan Cvetkovic,
To be or not to be is true. G. Boole No it isn't. L. E. J. Brouwer
!!! Sender/From address is bogus. Use reply-to one but edit it first!!!
Barbie LeVile
Rich Teer <rich...@rite-group.com> wrote:
>
> > it's cheap... they want to run it because it's free -- and I don't
> > mean the price... I mean that it's GPL code. Free as in freedom.
>
> If that's the case, why hasn't *BSD taken the world by storm?
> It's as free as the Linux kernel, and (most of) the userspace
> apps are the SAME.
>
Actualy, the BSd stuff is even more free, BSD license and all.
But linux people just scream louder. Just like MS and have the same goal.
--
--
Barbie - Prayers are like junkmail for Jesus
I have seen things you lusers would not believe.
I've seen Sun monitors on fire off the side of the multimedia lab.
I've seen NTU lights glitter in the dark near the Mail Gate.
All these things will be lost in time, like the root partition last week.
Time to die.
Paul Eggert
> He said 200 processors, not 64. Your benchmark wouldn't even get
> looked at; it fails the initial premise.
Yes, but Sun doesn't sell a 200-processor box either, so that premise
would exclude Sun as well.
It was reasonable to interpret his point to mean large numbers of
CPUs. 64-processor boxes are pretty large, by most people's measure.
Sun hasn't benchmarked anything larger than 72 processors on that
benchmark, but if you assume linear extrapolation the biggest box that
Sun sells (a 106 CPU Sun Fire 15K) would have about the same
performance on that benchmark as the 64 CPU SGI Altix running
GNU/Linux.
Rich Teer
> Actualy, the BSd stuff is even more free, BSD license and all.
> But linux people just scream louder. Just like MS and have the same goal.
Oh, I couldn't agree more.
Rich Teer
> It certainly doesn't get updated to keep up the pace with Linux
> developments.
Heh, even (some distributions of) Linux can't keep pace
with Linux developments! :-)
Paul Eggert
>> it's cheap... they want to run it because it's free -- and I don't
>> mean the price... I mean that it's GPL code. Free as in freedom.
>
> If that's the case, why hasn't *BSD taken the world by storm?
Partly it's because of what he said. The Linux kernel is GPLed; BSD
kernels are not. The GPL encourages people to give changes back to
the GNU/Linux developers, whereas the BSD folks are more often
hampered by outsiders keeping their changes private.
We use both BSD and Linux kernels in our commercial work here, and we
have gotten a good deal more help from the Linux community than the
BSD community, and this is why (other things being equal) we prefer
Linux to BSD kernels.
It of course was not inevitable that Linux kernels would be more
popular than BSD kernels, and the GPL isn't the only reason that Linux
is more popular; but it's one of the important reasons.
Andrew Gabriel
j...@cs.tu-berlin.de (Joerg Schilling) writes:
>
> This is why I like to see 'lkp' from SCO Unixware on Solaris.
For the benefit of those of us who haven't seen Unixware for
a long time, what does 'lkp' do?
--
Andrew Gabriel
Consultant Software Engineer
Joerg Schilling
Andrew Gabriel <and...@cucumber.demon.co.uk> wrote:
>In article <bkftq6$5ds$1...@news.cs.tu-berlin.de>,
> j...@cs.tu-berlin.de (Joerg Schilling) writes:
>>
>> This is why I like to see 'lkp' from SCO Unixware on Solaris.
>
>For the benefit of those of us who haven't seen Unixware for
>a long time, what does 'lkp' do?
It is Linux kernel support.
Lxrun just catches the signal that occurs when a Linux program tries to execute
a Linux syscall and emulates Linux in user space.
This gives only half of the possible speed....
lxrun is a kernel mudule that maps Linux syscalls inside the kernel into
SVr4 syscalls.
There is no noticable speed degradation - in many cases it is even faster than
on Linux.
If you call any Linux binary, you just get a root that is /linux from the
UnixWare wiew point. If you are running a Linux binary shell and execute a
UnixWare binary, this binary gets a root that is /unixware from the Linux
view point.
Philip Brown
>Rich Teer <rich...@rite-group.com> wrote in message news:<Pine.GSO.4.44.0309181442320.12004-100000@zaphod>...
>> $10 million SF 15K?
>
>This clearly shows how behind the times you guys are.
>
>The fastest Linux-based supercomputer ranks #3 on the top 500
>supercomputer list. The fastest Solaris-based supercomputer ranks
>#211 running on your precious "SF 15K". (http://www.top500.org/)
That isnt really a "supercomputer", it's a "linux cluster".
You cannot do everything you can do on a sunfire 15k, with a cluster of
smaller boxes.
"The system will include 1,058 eServer 325 systems with 2,116 AMD Opteron
processors."
1,058 systems != "_A_ supercomputer", reguardless of how the supercomputer
ranking list has been warped.
--
http://www.blastwave.org/ for solaris pre-packaged binaries with pkg-get
Organized by the author of pkg-get
[Trim the no-bots from my address to reply to me by email!]
S.1618 http://thomas.loc.gov/cgi-bin/bdquery/z?d105:SN01618:@@@D
http://www.spamlaws.com/state/ca1.html
Joerg Schilling
Paul Eggert <egg...@twinsun.com> wrote:
>At Fri, 19 Sep 2003 21:47:00 GMT, Rich Teer <rich...@rite-group.com> writes:
>
>>> it's cheap... they want to run it because it's free -- and I don't
>>> mean the price... I mean that it's GPL code. Free as in freedom.
>>
>> If that's the case, why hasn't *BSD taken the world by storm?
>
>Partly it's because of what he said. The Linux kernel is GPLed; BSD
>kernels are not. The GPL encourages people to give changes back to
>the GNU/Linux developers, whereas the BSD folks are more often
>hampered by outsiders keeping their changes private.
Then from a practical view, Linux is not GPLd :-(
Companies like RedHat and SuSE change programs like cdrecord and don't
even notify the Authors about their doings. The changes are (carefully
said) questionable, there seems to be no interest in a real cooperation.
The main reason for the changes seems to be to make programs like cdrecord
cooperate (from the belief of the RedHat/SuSE hackers) with their unusual
and hacked Linux kernel.
Reallitiy looks differrent, the hacked versions of cdrecord from RedHat
and SuSE are the cause of many "bug-reports" I receive via mail. This
is why I tell RedHat & SuSE GPL violators (see also GPL § 2 subclause c)
and GPL Preamble, subsection 6). The GPL request those unofficially and
hacked versions being marked in a way that users cleanly know that they
know that they are running an inofficial version. RedHat & SuSE don't contact
me; if they did thet would know that their hacks are bad.
The GPL seems to be a failed try now that the programs are used by commercial
acting companies. My impression about RedHat & SuSE is that they just pile up
what they can get for free and then start to make a proprietary version out
of it in hope that this method will tie cunstomers to their distribution...
>We use both BSD and Linux kernels in our commercial work here, and we
>have gotten a good deal more help from the Linux community than the
>BSD community, and this is why (other things being equal) we prefer
>Linux to BSD kernels.
BSD kernels are much more mature and give better stability in interfaces.
In contrary to the Linux developers, BSD developers do understand what
an interface is, so they can keep them stable..
>It of course was not inevitable that Linux kernels would be more
>popular than BSD kernels, and the GPL isn't the only reason that Linux
>is more popular; but it's one of the important reasons.
I still believe that it is only a Linux hype.
Richard L. Hamilton
de...@idirect.com (Derek Snider) writes:
[...]
> You're a little confused here. People don't want to run Linux because
> it's cheap... they want to run it because it's free -- and I don't
> mean the price... I mean that it's GPL code. Free as in freedom.
>
> Confused? This link will explain:
> http://www.gnu.org/philosophy/free-sw.html
Most people don't give a damn about philosophy; they want easier, better,
faster, and a bigger paycheck for doing less work.
Having said that, Sun's been fairly decent about making much of the source
available (at least to .edu's) to study, and has given away much stuff
outside of Solaris itself (lots of contributions to X, and OpenOffice, just
to name some major ones that come to mind). Ripping off prolly billions in
work just because not doing so doesn't fit your philosophy is a different
story entirely, and while I'd welcome their opening up more, I would
neither expect anything wholesale nor want them to give away all the
competitive advantages buried in there.
Information is _not_ free, it takes a bloody lot of work to come up with
useful information. One should be able to get paid for creating as well
as for maintaining...or give stuff away if one finds that to be more in
one's own self-interest. There is no friggin' common heritage of humanity
in all possible knowledge - any knowledge that can be used against
someone, or that they invested greatly to produce, should clearly be more
theirs than everyone else's, at least for awhile (although not forever).
There's a balance somewhere - IMO, the free bits everywhere radicals are
one kind of crazy, and the RIAA/MPAA/DMCA crowd are the opposite. They
both ought to have been spanked more often when they were little!
I welcome the BSD license, but I think the GPL (although not the LGPL) is
social activism run amok (although I freely accept that creators have the
choice to license as they please or not at all).
--
mailto:rlh...@mindwarp.smart.net http://www.smart.net/~rlhamil
Mr. Johan Andersson
On Sat, 20 Sep 2003, Philip Brown wrote:
> On 19 Sep 2003 12:03:43 -0700, de...@idirect.com wrote:
> >Rich Teer <rich...@rite-group.com> wrote in message news:<Pine.GSO.4.44.0309181442320.12004-100000@zaphod>...
> >> Who on EARTH would be stupid enough to run Linux on a
> >> $10 million SF 15K?
> >
> >This clearly shows how behind the times you guys are.
> >
> >The fastest Linux-based supercomputer ranks #3 on the top 500
> >supercomputer list. The fastest Solaris-based supercomputer ranks
> >#211 running on your precious "SF 15K". (http://www.top500.org/)
>
> That isnt really a "supercomputer", it's a "linux cluster".
> You cannot do everything you can do on a sunfire 15k, with a cluster of
> smaller boxes.
>
>
> "The system will include 1,058 eServer 325 systems with 2,116 AMD Opteron
> processors."
>
> 1,058 systems != "_A_ supercomputer", reguardless of how the supercomputer
> ranking list has been warped.
Just for comparasion, I dont know the numbers, but I am quite sure someone
here can do it, A local university here choose the StarCat SF15k over a
linux solution because while the linux solution probably could calculate
faster with a well written app, it by no way could mount the memory needed
to do the calculation. This SF15k has 72x900MHz and 280Gigabyte of RAM
all accesible for the data with 4 Terrabyte of disk connected, all in the
"box".
Its true the linux boxes would probably be faster, but I sure wouldnt like
to build the network behind it.
The SF15k we rolled in, installed OS and app, and ran in 36 hours.
Thats why some people choose the Sun Solution.
/Johan A
Chris Morgan
> On 19 Sep 2003 12:03:43 -0700, de...@idirect.com wrote:
> >Rich Teer <rich...@rite-group.com> wrote in message news:<Pine.GSO.4.44.0309181442320.12004-100000@zaphod>...
> >> Who on EARTH would be stupid enough to run Linux on a
> >> $10 million SF 15K?
> >
> >This clearly shows how behind the times you guys are.
> >
> >The fastest Linux-based supercomputer ranks #3 on the top 500
> >supercomputer list. The fastest Solaris-based supercomputer ranks
> >#211 running on your precious "SF 15K". (http://www.top500.org/)
>
> That isnt really a "supercomputer", it's a "linux cluster".
> You cannot do everything you can do on a sunfire 15k, with a cluster of
> smaller boxes.
>
>
> "The system will include 1,058 eServer 325 systems with 2,116 AMD Opteron
> processors."
>
> 1,058 systems != "_A_ supercomputer", reguardless of how the supercomputer
> ranking list has been warped.
This clinging to "single system image" definition supercomputers is
what killed Cray.
Sorry, anyone whose application is suitably parallel in nature
naturally moves to clusters of some kind and doesn't worry what to
call their equipment. The people who are left do need the much higher
processor<->processor bandwidth and lower latency of a 'real'
supercomputer, but that's a small niche and SGI is still strong in it
(not to mention the "new" Cray/Tera with the new X1 - up to 65536GB
RAM and 4096 cpus).
Chris
--
Chris Morgan <cm at mihalis.net> http://www.mihalis.net
Temp sig. - Enquire within
Frank Cusack
> Frank Cusack <fcu...@fcusack.com> wrote in message news:<m3y8wlk...@magma.savecore.net>...
>> to run on SPARC hardware.
>
> You're a little confused here. People don't want to run Linux because
> it's cheap... they want to run it because it's free -- and I don't
> mean the price... I mean that it's GPL code. Free as in freedom.
>
> Confused? This link will explain:
> http://www.gnu.org/philosophy/free-sw.html
I'm afraid it's you that's confused. Most people don't care. BUSINESSES
do not care. Businesses have no social agenda, and simply want to be as
efficient as possible. Linux on the desktop, even with it's high admin-
istration costs compared to Solaris, gives that efficiency. For most uses
(ie, non-engineering), Windows is even better.
/fc
Frank Cusack
> Companies like RedHat and SuSE change programs like cdrecord and don't
> even notify the Authors about their doings.
Yup, it's disgusting.
/fc
Frank Cusack
> At Fri, 19 Sep 2003 21:47:00 GMT, Rich Teer <rich...@rite-group.com> writes:
>
>>> it's cheap... they want to run it because it's free -- and I don't
>>> mean the price... I mean that it's GPL code. Free as in freedom.
>>
>> If that's the case, why hasn't *BSD taken the world by storm?
>
> Partly it's because of what he said. The Linux kernel is GPLed; BSD
> kernels are not. The GPL encourages people to give changes back to
> the GNU/Linux developers, whereas the BSD folks are more often
> hampered by outsiders keeping their changes private.
...
It's probably impossible to really say, but IMHO GPL has little to
nothing to do with it. ** NO ONE CARES ** and especially not
businesses. (Obviously, SOME people do care. Most of them got onto
Linux first, and cared second.) For proof, I need only point you to
the largest Linux company -- RedHat. If they cared about GPL, they
wouldn't push closed source proprietary software on people. Don't
know what I'm talking about? Do more research. RedHat doesn't care
about GPL, they are taking advantage of it. BTW, I think this is just
fine for a business to do -- their goal is to make money, not push a
social agenda. Getting back to my point, if people cared about GPL,
RH would not be the most popular distribution.
The reason for Linux's success (rather, popularity -- let's be clear) is
the bazaar development model. It will also be the reason for Linux's
downfall, should Sun get their act together.
Obviously, my opinion is colored by my own lack of interest in the
license wars. (I have published software under BSD, GPL and
commercial terms.)
/fc
Frank Cusack
friendly as Linux" comment:
On Fri, 19 Sep 2003 21:47:00 GMT Rich Teer <rich...@rite-group.com> wrote:
> On 19 Sep 2003, Derek Snider wrote:
>
>> There's no way that Solaris would ever be able to compete on that
>> issue.
>
> I agree - Sun would be foolish to opnely publish the Solaris
> source code.
I did not necessarily mean that (although I do advocate it). You can
number good open source kernel hackers in the dozens, or maybe a few
hundred. Most of those people won't care if Solaris is open sourced;
they will have no [psychological] ownership in Solaris and won't "convert".
I'm talking about encouraging "Solaris" development as opposed to "Linux"
development. You could think of it as a grass-roots marketing campaign.
It's a real shame that most Linux folks don't even know that Solaris is
markedly better -- and I'm not just talking about the kernel, I'm talking
about the system as a whole. Then again, most Linux folks are probably
only Linux folks because of it's popularity, not because they did any kind
of "competitive analysis". It's the popularity war that is going to kill
Sun.
/fc
Philip J. Koenig
writes...
IIRC, all the "top ranking supercomputers" for some years now have
been giant parallel beasts. IBM's "ASCII White" and "ASCII Blue"
AFAIK are basically clusters of POWER-chip boxes.
--
* Few people are capable of expressing with equanimity opinions which *
* differ from the prejudices of their social environment. Most people are *
* even incapable of forming such opinions. -- Albert Einstein *
* *
* To send email, remove numbers and spaces: pjkusenet64 @ ekahuna27 . com *
* Simple answers are for simple minds. Try a new way of looking at things. *
Paul Eggert
> Companies like RedHat and SuSE change programs like cdrecord and don't
> even notify the Authors about their doings.
The GPL does not require them to notify you, but if you're interested
in what they're doing it's easy for you to get copies of their source
code, if only to figure out a better way of doing it. This is in
contrast to the BSD model, where you won't even know what their code
does (and they certainly don't have to notify you). The GPL model
provides a huge advantage for you here, over the BSD model.
> My impression about RedHat & SuSE is that they just pile up what
> they can get for free and then start to make a proprietary version
> out of it in hope that this method will tie cunstomers to their
> distribution...
And what else did you expect them to do? (Sun does the same thing.)
This is one reason I prefer to use Debian GNU/Linux.
> BSD kernels are much more mature
This was certainly true in the past, and to some extent BSD kernels
are still more mature; but times are changing.
Paul Eggert
> I need only point you to the largest Linux company -- RedHat. If
> they cared about GPL, they wouldn't push closed source proprietary
> software on people.
If they didn't care about GPL, they'd be using BSD, no?
> The reason for Linux's success (rather, popularity -- let's be clear) is
> the bazaar development model.
... a model that BSD uses as well. So why is Linux doing so much better?
Joerg Schilling
Paul Eggert <egg...@twinsun.com> wrote:
>At 20 Sep 2003 09:14:04 GMT, j...@cs.tu-berlin.de (Joerg Schilling) writes:
>
>> Companies like RedHat and SuSE change programs like cdrecord and don't
>> even notify the Authors about their doings.
>
>The GPL does not require them to notify you, but if you're interested
>in what they're doing it's easy for you to get copies of their source
>code, if only to figure out a better way of doing it. This is in
>contrast to the BSD model, where you won't even know what their code
>does (and they certainly don't have to notify you). The GPL model
>provides a huge advantage for you here, over the BSD model.
I see not difference between the BSD model (where people and companies
may change the code legally without notice) and what RadHat and SuSE do
(where they hide the feact that they did change the code).
Fact is that they violate the GPL and that they ignore my mail where
I wrote them that they need to add a passus to the start up messages that
makes it clear that their version of cdrecord is not the original and thus
may have bugs not present in the original. I belive that is is even fair as
it is _known_ that there are bugs not present in the original.
>> My impression about RedHat & SuSE is that they just pile up what
>> they can get for free and then start to make a proprietary version
>> out of it in hope that this method will tie cunstomers to their
>> distribution...
>
>And what else did you expect them to do? (Sun does the same thing.)
>This is one reason I prefer to use Debian GNU/Linux.
The difference is that Sun does not state that they are a better company
because they donate back to the OSS comunity.
>> BSD kernels are much more mature
>
>This was certainly true in the past, and to some extent BSD kernels
>are still more mature; but times are changing.
I don't know what your main target is, but I recently came into a really bad
problem on Linux caused by the fact that Linux does VM overcommiting.The result
was a completely unstable Linux and it even looked like there was a multi
processor problem..... missing administrative and debugging programs
make it impossible to find the reason :-(
The other very important thing is SCSI. I am now writing for nearly 7 years that
Linux SCSI is not usable for SCSI target program (like cdrecord) development
because the kernel does not give back enough error information.
I even did write a patch that was intended to be the first step in a better but
up/downwards compatible SCSI user/kernel interface. This patch was prevented
to appear in the official Linux by Mr. Torvalds, Mr. Cox et. al. The reason was
obviously that they don't understand what an interface is to they don't
understand what to keep stable :-(
Linux kernel development uses the Polit Office Method. A few people that are
missing the needed knowledge decide on what happens and prevent better code to
appear.
The GPL definitely is not a grant for good/better software. Linux just gets
better propaganda.
Joerg Schilling
Frank Cusack <fcu...@fcusack.com> wrote:
>The reason for Linux's success (rather, popularity -- let's be clear) is
>the bazaar development model. It will also be the reason for Linux's
>downfall, should Sun get their act together.
The linux kernel uses the "polite office method": a few incapable people
prevent progress. It is a real wonder that there is some progress though.
Joerg Schilling
Paul Eggert <egg...@twinsun.com> wrote:
>If they didn't care about GPL, they'd be using BSD, no?
>
>> The reason for Linux's success (rather, popularity -- let's be clear) is
>> the bazaar development model.
>
>... a model that BSD uses as well. So why is Linux doing so much better?
Linux is only doing better in terms of propaganda.
Philip J. Koenig
writes...
> Look at Suns own systems. The Xeon based 1U and 2U systems are
> faster and cheaper that the UltraSPARC IIIi based alternatives.
It's really hard to compete with the economy of scale that the
PC market has. Intel makes gazillions of those CPUs, compared
to the few Sparcs that get built, for example. There is no way
that any 2nd-tier (according to volume) vendor can compete with
that. The only reason the IBM/Motorola architecture is even
remotely competitive (IMHO) is because they are pushing that
CPU into all sorts of other markets in order to keep the economy
of scale going. Beyond the Apple and IBM market, it goes into
various embedded systems of all kinds, automotive control, co-
processors on expansion cards, audio equipment, etc etc.
MIPS couldn't stay cost-competitive, Alpha couldn't stay cost-
competitive, PA-RISC, etc. No one but the PowerPC and x86 are
really in the running for the mass market any more, and even
in the middle-market that same economy of scale brings down the
cost of server-level components.
When you can't tap into those CPU and peripheral production
efficiencies, you really can't compete anywhere but the high end.
Both Sun and Apple have abandoned SCSI drives in a lot of hardware
for that reason, but there are many other areas where Sun can't
or won't do that. (I think the last thing Sun wants to do is be
reduced to just another PC cloner)
Opteron could be an interesting angle, particularly if they
engineer some Sun-exclusive enhancements that other companies
don't offer.
Philip J. Koenig
LeVile) writes...
> On Fri, 19 Sep 2003 21:47:00 GMT
> Rich Teer <rich...@rite-group.com> wrote:
>
>
> >
> > > it's cheap... they want to run it because it's free -- and I don't
> > > mean the price... I mean that it's GPL code. Free as in freedom.
> >
> > If that's the case, why hasn't *BSD taken the world by storm?
> > It's as free as the Linux kernel, and (most of) the userspace
> > apps are the SAME.
> >
>
> Actualy, the BSd stuff is even more free, BSD license and all.
> But linux people just scream louder. Just like MS and have the same goal.
Not really. There are a lot more people working on Linux than
there are on the open-source BSD variants. Thus you get new
drivers faster, you get lots of interesting features quicker,
you have a bigger community to debug things, etc. For example,
nowadays there are no less than 3 or 4 different journaling
filesystems you can use on a Linux box. On FreeBSD/OpenBSD/
NetBSD there are zero, AFAIK. (working on it, but nothing shipping
yet)
I'd really love it if a significant portion of the people hacking
Linux would switch over to ie FreeBSD, because in many ways I like
the development model a lot better - none of this Linux distro
chaos. But that's not happening any time soon.
Thomas Dehn
"Philip J. Koenig" <See_email_@ddress_below.This_one_is.invalid> wrote:
> > > >The fastest Linux-based supercomputer ranks #3 on the top 500
> > > >supercomputer list. The fastest Solaris-based supercomputer ranks
> > > >#211 running on your precious "SF 15K". (http://www.top500.org/)
> > >
> > > That isnt really a "supercomputer", it's a "linux cluster".
> > > You cannot do everything you can do on a sunfire 15k, with a cluster of
> > > smaller boxes.
> IIRC, all the "top ranking supercomputers" for some years now have
> been giant parallel beasts. IBM's "ASCII White" and "ASCII Blue"
> AFAIK are basically clusters of POWER-chip boxes.
The Earth Simulator is just a huge single box.
The problem with "top ranking supercomputer" charts
is that they are misleading for several reasons
- large commercial installations usually cannot afford
the downtime to run the benchmark, or do not want to
participate for security reasons, and as a result
they are not listed in these rankings (except for those few where
single boxes still make the list).
- the benchmarks measure GigaFlops, which
is a useless number. A typical distributed parallel
algorithm will have to do considerable extra work
compared to the fastest serial algorithm, sometimes up to the
point where, say, execution on 64 CPUs takes longer
than execution on 32 CPUs. Counting such wasted extra CPU
cycles as additional GigaFlops is nonsense.
- the benchmarks do not measure sustained throughput
under real life conditions. And, if you have to adapt your
code to make it run fast on your new cluster, the cycles
necessary to migrate your code should be subtracted
from the system's performance.
Thomas
Rich Teer
> I'm talking about encouraging "Solaris" development as opposed to "Linux"
> development. You could think of it as a grass-roots marketing campaign.
>
> It's a real shame that most Linux folks don't even know that Solaris is
> markedly better -- and I'm not just talking about the kernel, I'm talking
> about the system as a whole. Then again, most Linux folks are probably
> only Linux folks because of it's popularity, not because they did any kind
> of "competitive analysis". It's the popularity war that is going to kill
> Sun.
I think we agree about 100% on this!
Frank Cusack
> In article <m3vfrmq...@magma.savecore.net>,
> Frank Cusack <fcu...@fcusack.com> wrote:
>
>>The reason for Linux's success (rather, popularity -- let's be clear) is
>>the bazaar development model. It will also be the reason for Linux's
>>downfall, should Sun get their act together.
>
>
> The linux kernel uses the "polite office method": a few incapable people
> prevent progress. It is a real wonder that there is some progress though.
"Linux" is not just the kernel. The kernel itself is less of a bazaar
than BSD is.
/fc
Paul Floyd
<See_email_@ddress_below.This_one_is.invalid> wrote:
[snip]
> you have a bigger community to debug things, etc. For example,
> nowadays there are no less than 3 or 4 different journaling
> filesystems you can use on a Linux box. On FreeBSD/OpenBSD/
> NetBSD there are zero, AFAIK. (working on it, but nothing shipping
> yet)
Well, personally I'd rather have one top-notch journalling FS than a
multitude of mediocre ones. [Note: I've never used a journalled FS on
Linux, but for other apps, a plethora of bad ones seems to be the norm].
A bientot
Paul
--
Paul Floyd http://paulf.free.fr (for what it's worth)
Surgery: ennobled Gerald.
Frank Cusack
No, BSD is a complete system and is closer to a cathedral model (core team).
Linux is ad-hoc. Note that I'm talking about the system as a whole, the
"operating environment" in Sun lingo.
The ability and encouragement of many variations allows and EMPOWERS people
to create their own distributions, and to contribute more strongly. Sun
et al. have helped fuel this by not including GNU tools as a core part of
their OS. That has FORCED the "professional unix users" to piece together
their systems, Linux style. (Let's look at Solaris: bash didn't show up
until Solaris 8!) The SysV packaging system makes this worse. That's
a complete essay in itself, I won't bore you.
The ability for a single developer, or even someone with a bug fix, to get
their patch into Linux has fewer obstacles than BSD. With Linux, if you
want something to work your way, and meet resistance, it's pretty easy for
you to just become the maintainer. If it's a good idea, people will come
to you. With BSD, power isn't distributed that way.
Poor software quality has contributed to Linux's success as well. It is
now ingrained that all software sucks. So people are willing to accept
the problems of Linux in exchange for the accessibility.
/fc
Frank Cusack
> Not really. There are a lot more people working on Linux than
> there are on the open-source BSD variants. Thus you get new
> drivers faster, you get lots of interesting features quicker,
> you have a bigger community to debug things, etc. For example,
> nowadays there are no less than 3 or 4 different journaling
> filesystems you can use on a Linux box. On FreeBSD/OpenBSD/
> NetBSD there are zero, AFAIK. (working on it, but nothing shipping
> yet)
True, there are zero journalling filesystems, but you have something
better: softupdates. LFS is continually worked on, other than that
no one is working on journalling AFAIK ... and why bother? softupdates
is already there.
(LFS is like journalling.)
That's not to detract from your point that many more people work on
Linux, this fuels it's popularity and -- double-whammy -- takes away
from BSD's popularity.
/fc
Dave Uhring
> "Linux" is not just the kernel.
Really? What else is it then?
Frank Cusack
The tools that go with it.
If you have JUST the kernel, you have nothing. It's the sum of all things
that come with it.
I just meant that in common usage, people saying "Linux" are not referring
to just the kernel.
/fc
Dave Uhring
You make Richard Stallman unhappy.
http://www.gnu.org/gnu/linux-and-gnu.html
"There really is a Linux, and these people are using it, but it is not the
operating system. Linux is the kernel: the program in the system that
allocates the machine's resources to the other programs that you run."
Perhaps you meant a "Linux distribution", or to make RMS happier a
"GNU/Linux distribution".
Joerg Schilling
Philip J. Koenig <See_email_@ddress_below.This_one_is.invalid> wrote:
>Not really. There are a lot more people working on Linux than
>there are on the open-source BSD variants. Thus you get new
>drivers faster, you get lots of interesting features quicker,
>you have a bigger community to debug things, etc. For example,
>nowadays there are no less than 3 or 4 different journaling
>filesystems you can use on a Linux box. On FreeBSD/OpenBSD/
>NetBSD there are zero, AFAIK. (working on it, but nothing shipping
>yet)
You don't need a JFS on BSD because there are inode soft updates.
Ask Kirk McKusick if you don't believe it...
Linux has a lot 80% file system implementations; that doesn't make
Linux better.
A big problem with filesystems on Linux is that they are fast only because
they disregard filesyetem integrity. Once you try to write software that
does try to keep the filesystem state 'known', you get a performance degradation
by 400% on Linux. If you don't believe me, compare 'star' and 'GNU tar'
extraction speed for a Linux kernel tar ball. GNU tar is 4x faster than
'star' unless you call 'star -no-fsync ...'. If you do the same test
on Solaris, you get only 20% win with -no-fsync although the Solaris UFS
cares about filesystem integrity.
The funny thing is that Linux followers even like me to make star as unable to
detect write errors as GNU tar is :-(
Logan Shaw
> On Sun, 21 Sep 2003 15:14:36 -0700, Frank Cusack wrote:
>>I just meant that in common usage, people saying "Linux" are not referring
>>to just the kernel.
> You make Richard Stallman unhappy.
I suppose so, and I even understand why he might feel that way.
But I'm not sure that is enough to stop me from calling the whole
collection "Linux".
If anything, I'd prefer to call it "Linux/GNU" rather than "Linux",
because not all of the userland stuff is GNU after all...
- Logan
Logan Shaw
> If anything, I'd prefer to call it "Linux/GNU" rather than "Linux",
> because not all of the userland stuff is GNU after all...
Oops, that should say '"Linux/GNU" rather than "GNU/Linux"'.
- Logan
Dave Uhring
Actually, IMHO, the vanity of Stallman is somewhat excessive. Much less
than 50% of a typical Linux distribution consists of GNU software.
Philip J. Koenig
writes...
> In article <MPG.19d71bcc1...@corp.supernews.com>,
> Philip J. Koenig <See_email_@ddress_below.This_one_is.invalid> wrote:
>
> >Not really. There are a lot more people working on Linux than
> >there are on the open-source BSD variants. Thus you get new
> >drivers faster, you get lots of interesting features quicker,
> >you have a bigger community to debug things, etc. For example,
> >nowadays there are no less than 3 or 4 different journaling
> >filesystems you can use on a Linux box. On FreeBSD/OpenBSD/
> >NetBSD there are zero, AFAIK. (working on it, but nothing shipping
> >yet)
>
> You don't need a JFS on BSD because there are inode soft updates.
> Ask Kirk McKusick if you don't believe it...
I know about soft updates but I was under the impression that it
only protects meta-data, not the data itself, and it doesn't help
reduce the amount of time it takes to do an fsck on a dirty volume.
(although FreeBSD has some kind of "fast fsck" thingy that I think
they added in the last 6 months or so)
> Linux has a lot 80% file system implementations; that doesn't make
> Linux better.
>
> A big problem with filesystems on Linux is that they are fast only because
> they disregard filesyetem integrity. Once you try to write software that
> does try to keep the filesystem state 'known', you get a performance degradation
> by 400% on Linux. If you don't believe me, compare 'star' and 'GNU tar'
> extraction speed for a Linux kernel tar ball. GNU tar is 4x faster than
> 'star' unless you call 'star -no-fsync ...'. If you do the same test
> on Solaris, you get only 20% win with -no-fsync although the Solaris UFS
> cares about filesystem integrity.
>
> The funny thing is that Linux followers even like me to make star as unable to
> detect write errors as GNU tar is :-(
Doesn't surprise me. I'm actually really new to Solaris from an admin
point of view (had a number of accounts on SunOS boxes at ISPs over
the years but never administered one) and one of the things that
struck me so far in my extremely recent Solaris admin experience is the
depth and complexity compared to the open source variants. I think the
open-source crowd is "hitting the high points", but a lot of the less-
glamorous details (filesystem integrity, large system management, etc)
seem to get either de-prioritized or not done at all.
With BSD in many cases it seems the problem is simply lack of manpower,
with Linux I think it's a matter of doing what's interesting rather
than what might be important (seems to be the common problem when people
are doing things for pleasure rather than being paid to do them) although
in the Linux case there are so many people contributing that things are
definitely improving very fast in many areas.
Rich Teer
> The tools that go with it.
>
> If you have JUST the kernel, you have nothing. It's the sum of all things
> that come with it.
>
> I just meant that in common usage, people saying "Linux" are not referring
> to just the kernel.
Many (most?) of those other things are equally bundled with
the BSDs presumably, so they don't add to "Linux" specifically.
Paul Eggert
> Fact is that they violate the GPL and that they ignore my mail
If that's true, then sue them. If you have a case, you'll win.
If your code had been been distributed under the BSD license,
you wouldn't have had a case, so the GPL is a win here.
> Sun does not state that they are a better company
> because they donate back to the OSS comunity.
Sure they do. They've done so on many occasions. OpenOffice is one
example, and Sun has cited it as giveback, and rightly so.
> A few people that are missing the needed knowledge decide on what
> happens and prevent better code to appear.
You're always free to build a better distribution, based on their
work. This has happened many times to GNU projects (including GCC and
Emacs, both core projects). I've even done it myself once or twice,
and I've had it done to me once. It's a healthy process. You can't
always sit back and expect every maintainer to agree with you;
sometimes you have to do the work and distribute the result yourself.
Paul Eggert
> Linux is only doing better in terms of propaganda.
That's certainly not true. Linux kernels outperform BSD kernels in
many areas. I'll give just one example: the SPECint_rate_base2000
benchmark. The best published GNU/Linux score is 601 (the SGI Altix,
with 64 1.3 GHz Itanium 2 processors). This far exceeds any published
BSD score.
Reference:
http://www.specbench.org/cpu2000/results/res2003q3/cpu2000-20030616-02239.html
Philip Brown
>phi...@bolthole.no-bots.com (Philip Brown) writes:
>> That isnt really a "supercomputer", it's a "linux cluster".
>> You cannot do everything you can do on a sunfire 15k, with a cluster of
>> smaller boxes.
>>
>> ...
>what killed Cray.
>
>Sorry, anyone whose application is suitably parallel in nature
>naturally moves to clusters of some kind and doesn't worry what to
>call their equipment.
theyn why do they keep trying to rename their systems "supercomputer"
instead of "cluster"?
Obviously, it DOES matter what the equipment is called.
--
http://www.blastwave.org/ for solaris pre-packaged binaries with pkg-get
Organized by the author of pkg-get
[Trim the no-bots from my address to reply to me by email!]
S.1618 http://thomas.loc.gov/cgi-bin/bdquery/z?d105:SN01618:@@@D
http://www.spamlaws.com/state/ca1.html
Joerg Schilling
Logan Shaw <lshaw-...@austin.rr.com> wrote:
>> You make Richard Stallman unhappy.
He tries to get the wrong credit....
>I suppose so, and I even understand why he might feel that way.
>But I'm not sure that is enough to stop me from calling the whole
>collection "Linux".
>
>If anything, I'd prefer to call it "Linux/GNU" rather than "Linux",
>because not all of the userland stuff is GNU after all...
Not to forget that there is X and a lot of software that is not initiated
by the FSF. It is a really bad idea to call something with a name that starts
with "GNU" when less than 50% of it is FSF initiated.
Alessandro Selli
|
|"Philip J. Koenig" <See_email_@ddress_below.This_one_is.invalid> wrote:
|> > > >The fastest Linux-based supercomputer ranks #3 on the top 500
|> > > >supercomputer list. The fastest Solaris-based supercomputer ranks
|> > > >#211 running on your precious "SF 15K". (http://www.top500.org/)
|> > >
|> > > That isnt really a "supercomputer", it's a "linux cluster".
|> > > You cannot do everything you can do on a sunfire 15k, with a cluster of
|> > > smaller boxes.
|[...]
|> IIRC, all the "top ranking supercomputers" for some years now have
|> been giant parallel beasts. IBM's "ASCII White" and "ASCII Blue"
|> AFAIK are basically clusters of POWER-chip boxes.
|
|The Earth Simulator is just a huge single box.
Now, this is not my field of competence, but:
http://www.es.jamstec.go.jp/esc/eng/Hardware/system.html
System Configuration
The ES is a highly parallel vector supercomputer system of the
distributed-memory type, and consisted of 640 processor nodes (PNs) connected
by 640x640 single-stage crossbar switches. Each PN is a system with a shared
memory, consisting of 8 vector-type arithmetic processors (APs), a 16-GB main
memory system (MS), a remote access control unit (RCU), and an I/O processor.
The peak performance of each AP is 8Gflops. The ES as a whole thus consists of
5120 APs with 10 TB of main memory and the theoretical performance of 40Tflops
The pictures do seem to indicate a cluster system with an "Interconnection
Network" as its core.
Sandro
--
Bellum se ipsum alet
La guerra nutre se stessa
Livio, Ab urbe condita, XXXIV,9
Alessandro Selli
|de...@idirect.com (Derek Snider) writes in comp.unix.solaris:
||IBM's AIX 5L allows for source-level Linux compatability whereby
||making cross-platform compatable code a breeze. Code designed to be
||built under Linux will compile under AIX 5L without a hitch.
|
|That's truly amazing since much Linux code won't even build under
|different versions/distros of Linux.
Let me remind you, Linux is a kernel. You are talking of applications, here.
I could compile the SSH 3.7.1p1 and SSL 0.9.7b on three years old
distributions. What similar experiences did you have?
Why doesn't Solaris9 support my Sparc5 sound hardware?
Alessandro Selli
[...]
|> it's cheap... they want to run it because it's free -- and I don't
|> mean the price... I mean that it's GPL code. Free as in freedom.
|
|If that's the case, why hasn't *BSD taken the world by storm?
|It's as free as the Linux kernel, and (most of) the userspace
|apps are the SAME.
http://www.gnu.org/philosophy/bsd.html
[...]
Most of them are equivalent except for details of wording, but the license
used for BSD until 1999 had a special problem: the `obnoxious BSD advertising
clause''. It said that every advertisement mentioning the software must
include a particular sentence:
3. All advertising materials mentioning features or use of this software
must display the following acknowledgement:
This product includes software developed by the University of
California, Berkeley and its contributors.
Initially the obnoxious BSD advertising clause was used only in the Berkeley
Software Distribution. That did not cause any particular problem, because
including one sentence in an ad is not a great practical difficulty.
If other developers who used BSD-like licenses had copied the BSD advertising
clause verbatim--including the sentence that refers to the University of
California--then they would not have made the problem any bigger.
But, as you might expect, other developers did not copy the clause verbatim.
They changed it, replacing `University of California'' with their own
institution or their own names. The result is a plethora of licenses,
requiring a plethora of different sentences.
When people put many such programs together in an operating system, the result
is a serious problem. Imagine if a software system required 75 different
sentences, each one naming a different author or group of authors. To
advertise that, you would need a full-page ad.
This might seem like extrapolation ad absurdum, but it is actual fact. NetBSD
comes with a long list of different sentences, required by the various
licenses for parts of the system. In a 1997 version of NetBSD, I counted 75 of
these sentences. I would not be surprised if the list has grown by now.
[et cetera]
Alessandro Selli
|In article <7wk784j...@twinsun.com>,
|Paul Eggert <egg...@twinsun.com> wrote:
|>At Fri, 19 Sep 2003 21:47:00 GMT, Rich Teer <rich...@rite-group.com> writes:
|>
|>>> it's cheap... they want to run it because it's free -- and I don't
|>>> mean the price... I mean that it's GPL code. Free as in freedom.
|>>
|>> If that's the case, why hasn't *BSD taken the world by storm?
|>
|>Partly it's because of what he said. The Linux kernel is GPLed; BSD
|>kernels are not. The GPL encourages people to give changes back to
|>the GNU/Linux developers, whereas the BSD folks are more often
|>hampered by outsiders keeping their changes private.
|
|Then from a practical view, Linux is not GPLd :-(
|
|Companies like RedHat and SuSE change programs like cdrecord and don't
|even notify the Authors about their doings. The changes are (carefully
|said) questionable, there seems to be no interest in a real cooperation.
|The main reason for the changes seems to be to make programs like cdrecord
|cooperate (from the belief of the RedHat/SuSE hackers) with their unusual
|and hacked Linux kernel.
[...]
This was a much debated point: should the GPL have authors of changes notify
the author of the original code they produced those changes? They eventually
decided that nofifies where not to be compulsory under the GPL, mainly because
in some occasions the author of the changes would not be able to communicate
his changes to the author of the original code (this would happen in many
african or asian places where they use - and develop - GPL code but still lack
internet connectivity or can have it at an expence they cannot afford).
Alessandro Selli
|It's probably impossible to really say, but IMHO GPL has little to
|nothing to do with it. ** NO ONE CARES ** and especially not
|businesses. (Obviously, SOME people do care. Most of them got onto
|Linux first, and cared second.) For proof, I need only point you to
|the largest Linux company -- RedHat. If they cared about GPL, they
|wouldn't push closed source proprietary software on people. Don't
|know what I'm talking about? Do more research. RedHat doesn't care
|about GPL, they are taking advantage of it.
If this was true, then why did RedHat release their own installation
software (Anaconda) as GPL code (as opposed to SuSE, their YAST2 [Yet Another
Setup Tool] is proprietary)? And why did they do the same with the
software they developped that detects the hardware on the system and
configures the kernel drivers accordingly (kudzu)? Even the hardware database
kudzu uses is GPL! Beginning with the current stable release, Debian too
carries kudzu.
Alan Coopersmith
|Il giorno Fri, 19 Sep 2003, Alan Coopersmith cosě ha scritto:
|
||de...@idirect.com (Derek Snider) writes in comp.unix.solaris:
|||IBM's AIX 5L allows for source-level Linux compatability whereby
|||making cross-platform compatable code a breeze. Code designed to be
|||built under Linux will compile under AIX 5L without a hitch.
||
||That's truly amazing since much Linux code won't even build under
||different versions/distros of Linux.
|
| Let me remind you, Linux is a kernel. You are talking of applications, here.
So was the original poster, since that was my understanding of the
"Linux compatibility" offered by AIX 5L, not kernel module
compatibility.
--
________________________________________________________________________
Alan Coopersmith al...@alum.calberkeley.org
http://www.CSUA.Berkeley.EDU/~alanc/ aka: Alan.Coo...@Sun.COM
Working for, but definitely not speaking for, Sun Microsystems, Inc.
Chris Morgan
> >Sorry, anyone whose application is suitably parallel in nature
> >naturally moves to clusters of some kind and doesn't worry what to
> >call their equipment.
>
> theyn why do they keep trying to rename their systems "supercomputer"
> instead of "cluster"?
> Obviously, it DOES matter what the equipment is called.
Research grants, ego, habit, bias, nostalgia. Take your pick.
--
Chris Morgan
"Post posting of policy changes by the boss will result in
real rule revisions that are irreversible"
- anonymous correspondent
Chris Morgan
> It's a real shame that most Linux folks don't even know that Solaris is
> markedly better -- and I'm not just talking about the kernel, I'm talking
> about the system as a whole. Then again, most Linux folks are probably
> only Linux folks because of it's popularity, not because they did any kind
> of "competitive analysis". It's the popularity war that is going to kill
> Sun.
I'm not "linux folk" but I do run both Linux and Solaris systems.
I don't think the contention that Solaris is markedly better can stand
without further qualification. Unless we're indulging in pure
knee-jerk tribalism.
As a internet-facing webserver, I prefer Linux because of
price/performance and rapidity of patches for security
problems. Usually same day (e.g. the recent OpenSSH mess).
For a development platform, I kind of prefer Solaris, except when I'm
working with FOSS packages (see next point) in which case it's the
other way around.
For a graphics workstation, I now prefer a Linux box for a number of
reasons that I have probably talked about too much already on this
newsgroup.
As far as the system as a whole, Solaris lags behind some of the Linux
distros in terms of FOSS it comes with, but wins on stability. I've
heard people claim that Solaris is way ahead on SMP machines once the
number of cpus is large, however I can't afford such machines to run
any operating system on, and anyway I've heard other people deny
(e.g. SGI Altix machine runs Linux).
Another claim is that Sun wins because of the support, however I've
had Sun support, including people come to my desk and open up my Blade
1000 and do stuff to it, and quite frankly it has been mixed. And the
'support' that came with my v240, which I ordered in April and is
still not in production wasn't worth a cent.
Chris
Chris Morgan
there's a great quote from someone on this topic in the book "When
Wizards Stay Up Late" which is about the history of the
internet. Something along the lines of "each person comes along and
adds a brick and says 'I built a cathedral'"
The best I could find on the net is this (bad translation of german
translation of original)
Frank Cusack
> Il giorno Sat, 20 Sep 2003, Frank Cusack cosě ha scritto:
>
> |It's probably impossible to really say, but IMHO GPL has little to
> |nothing to do with it. ** NO ONE CARES ** and especially not
> |businesses. (Obviously, SOME people do care. Most of them got onto
> |Linux first, and cared second.) For proof, I need only point you to
> |the largest Linux company -- RedHat. If they cared about GPL, they
> |wouldn't push closed source proprietary software on people. Don't
> |know what I'm talking about? Do more research. RedHat doesn't care
> |about GPL, they are taking advantage of it.
>
> If this was true, then why did RedHat release their own
> installation software (Anaconda) as GPL code (as opposed to SuSE,
> their YAST2 [Yet Another Setup Tool] is proprietary)? And why did
> they do the same with the software they developped that detects the
> hardware on the system and configures the kernel drivers accordingly
> (kudzu)? Even the hardware database kudzu uses is GPL! Beginning
> with the current stable release, Debian too carries kudzu.
Because a) they didn't consider at the time that it had any revenue
generating value to them and b) they SHIP the software with the system.
They want the appearance of shipping GPL software.
Why doesn't RH GPL the update daemon? Because they will lose revenue
if everyone can run their own.
/fc
Dave Uhring
> there's a great quote from someone on this topic in the book "When
> Wizards Stay Up Late" which is about the history of the
> internet. Something along the lines of "each person comes along and
> adds a brick and says 'I built a cathedral'"
There are 408 'bricks' installed on my Slackware-9.0 system. Few of them
originated at the FSF.