Why is SUN falling so far behind IBM?
Mike Cox
technology, dual CPUs that if one fails, the other takes over, a
service manager, and reduncant powersupplies and RAID disks. Best of
all, that server will work seemlessly with my Linux desktops and
applications because it too runs linux.
SUN is going in the complete opposite direction. It is blindly
clinging on to Solaris instead of switching to Linux OR having a Linux
affinity Solaris such as something that could be branded "SolarisLX"
that would work seemlessly with Linux. Make Linux or Linux affinity
Solaris availible for everything from a SUNFIRE to the Ultrasparc.
The last thing that SUN is doing wrong is ruining the SUN brand by
using commodity INTEL chips. Companies used to show their customers
and investors the IT room full of huge SUN boxes to impress them, but
now that the SUN brand has been tarnished it is not as impressive
anymore. Customers now see sun as either a has-been or just another
OEM DELL machine.
SUN was high-end reliable computing that boasted you could write your
app on an Ultra and move it to a SUNFIRE. IBM now has that message
because now you can write your app on Linux and run it on a Mainframe.
Fun too how much money IBM makes from JAVA.
Gregory Toomey
> For about 12,000 USD, you can buy a pSeries that has ChipKill(TM)
> technology, dual CPUs that if one fails, the other takes over, a
> service manager, and reduncant powersupplies and RAID disks. Best of
> all, that server will work seemlessly with my Linux desktops and
> applications because it too runs linux.
>
> SUN is going in the complete opposite direction. It is blindly
> clinging on to Solaris instead of switching to Linux
Sun owns Cobalt, which makes linux servers.
> OR having a Linux
> affinity Solaris such as something that could be branded "SolarisLX"
> that would work seemlessly with Linux. Make Linux or Linux affinity
> Solaris availible for everything from a SUNFIRE to the Ultrasparc.
>
> The last thing that SUN is doing wrong is ruining the SUN brand by
> using commodity INTEL chips. Companies used to show their customers
> and investors the IT room full of huge SUN boxes to impress them, but
> now that the SUN brand has been tarnished it is not as impressive
> anymore. Customers now see sun as either a has-been or just another
> OEM DELL machine.
In the past sun had to comete with IBM & HP. Sun now has to compete with the
$0 (or near to $0) linux alternative.
> SUN was high-end reliable computing that boasted you could write your
> app on an Ultra and move it to a SUNFIRE. IBM now has that message
> because now you can write your app on Linux and run it on a Mainframe.
> Fun too how much money IBM makes from JAVA.
Where big money is concerned (eg banks), buying a "solution", reliability
and service is more important than cost. Sun still makes lots of money
selling services.
But you are right in Sun & solaris getting squeezed. Linux & bsd lack some
functionality of solaris but cost near $0 . Its what happens when operating
systems become commodities.
gtoomey
CJT
> For about 12,000 USD, you can buy a pSeries that has ChipKill(TM)
> technology, dual CPUs that if one fails, the other takes over, a
> service manager, and reduncant powersupplies and RAID disks. Best of
> all, that server will work seemlessly with my Linux desktops and
> applications because it too runs linux.
>
<snip>
$12K can buy some pretty nice Sun gear, too.
--
The e-mail address in our reply-to line is reversed in an attempt to
minimize spam. Our true address is of the form che...@prodigy.net.
Scrubbing Bubbles
> For about 12,000 USD, you can buy a pSeries that has ChipKill(TM)
> technology, dual CPUs that if one fails, the other takes over, a
> service manager, and reduncant powersupplies and RAID disks. Best of
> all, that server will work seemlessly with my Linux desktops and
> applications because it too runs linux.
>
> SUN is going in the complete opposite direction. It is blindly
> clinging on to Solaris instead of switching to Linux OR having a Linux
Huh? They see the need for a high end Unix OS that is proprietary. Good
for them. Diversity is good.
> affinity Solaris such as something that could be branded "SolarisLX"
> that would work seemlessly with Linux. Make Linux or Linux affinity
> Solaris availible for everything from a SUNFIRE to the Ultrasparc.
Why ?
> The last thing that SUN is doing wrong is ruining the SUN brand by
> using commodity INTEL chips. Companies used to show their customers
That's only for cheap workstations. They need cheap JDS stations to be a
full service company. They get clobbered when they try to 'work together'
with deceitful people like Micro$haft, so, now they have a fully complete
system. They can do cheap workstations, servers, Linux, java, desktop,
high performance computing, Unix *and* Linux. And, of course, they own
java.
> and investors the IT room full of huge SUN boxes to impress them, but
> now that the SUN brand has been tarnished it is not as impressive
> anymore. Customers now see sun as either a has-been or just another
> OEM DELL machine.
Old news. This is the year 2004, Rip van Winkle. Read on:
Infoimaging
Sun Posts Modest Sales Gains (Folo) STMicro Profit Nearly Doubles As Chief
Forecasts Strong Demand
The New York Times, 07.23.04, 5:00 PM ET
Sun, the Santa Clara, California, computer hardware and software company,
said its revenue for the quarter that ended on June 30 was $3.11 billion,
up 4.3 percent.
Net income was $795 million, against a loss of $1.04 billion a year
earlier. But the figure for the most recent quarter included $1.6 billion
that Microsoft paid to settle a suit. Without special gains and charges,
Sun lost $169 million.
International Herald Tribune
(C) 2004 International Herald Tribune. via ProQuest Information and
Learning Company; All Rights Reserved
Rich Teer
[Pro-Linux drivel snipped.]
___________________
/| /| | |
||__|| | Please do |
/ O O\__ NOT |
/ \ feed the |
/ \ \ trolls |
/ _ \ \ ______________|
/ |\____\ \ ||
/ | | | |\____/ ||
/ \|_|_|/ \ __||
/ / \ |____| ||
/ | | /| | --|
| | |// |____ --|
* _ | |_|_|_| | \-/
*-- _--\ _ \ // |
/ _ \\ _ // | /
* / \_ /- | - | |
* ___ c_c_c_C/ \C_c_c_c____________
--
Rich Teer, SCNA, SCSA
President,
Rite Online Inc.
Voice: +1 (250) 979-1638
URL: http://www.rite-online.net
Gavin Maltby
> ___________________
> /| /| | |
> ||__|| | Please do |
> / O O\__ NOT |
> / \ feed the |
> / \ \ trolls |
> / _ \ \ ______________|
> / |\____\ \ ||
> / | | | |\____/ ||
> / \|_|_|/ \ __||
> / / \ |____| ||
> / | | /| | --|
> | | |// |____ --|
> * _ | |_|_|_| | \-/
> *-- _--\ _ \ // |
> / _ \\ _ // | /
> * / \_ /- | - | |
> * ___ c_c_c_C/ \C_c_c_c____________
>
Hey Rich - the grin looks remarkably as I remember you!
Gavin
Chris Cox
> For about 12,000 USD, you can buy a pSeries that has ChipKill(TM)
> technology, dual CPUs that if one fails, the other takes over, a
> service manager, and reduncant powersupplies and RAID disks. Best of
> all, that server will work seemlessly with my Linux desktops and
> applications because it too runs linux.
>
> SUN is going in the complete opposite direction. It is blindly
> clinging on to Solaris instead of switching to Linux OR having a Linux
> affinity Solaris such as something that could be branded "SolarisLX"
> that would work seemlessly with Linux. Make Linux or Linux affinity
> Solaris availible for everything from a SUNFIRE to the Ultrasparc.
Actually... I have word from the inside of Sun that Sun is "full tilt"
favoring Linux right now. Solaris will be an additional offering,
but after it is completely open sourced, users will have a choice
between FULLY Sun supported LINUX (e.g. JDS - which is built on
Novell SUSE Linux Enterprise Server v8), or FULLY Sun supported
Solaris.
Your information is wrong. Though many in the newsgroup will likely
support the idea of Sun NOT going after Linux, my inside sources
say that Linux has become king at Sun currently.
Where's IBM's Linux desktop offering? While Sun has the Java
Desktop System, IBM pretty much only support Linux on the server
side. Sun is actively pushing both server and client side Linux.
>
> The last thing that SUN is doing wrong is ruining the SUN brand by
> using commodity INTEL chips. Companies used to show their customers
> and investors the IT room full of huge SUN boxes to impress them, but
> now that the SUN brand has been tarnished it is not as impressive
> anymore. Customers now see sun as either a has-been or just another
> OEM DELL machine.
You'll find the Opteron a better price/performance target than
the IBM offerings currently.... saying that, IBM also offers an
Opteron line... so I guess maybe I don't fully understand the
issue here. Sparc, will live on for a while.. possibly even with
another upgrade or two... but you don't have to use Sparc anymore,
and IMHO, that's a good thing.
>
> SUN was high-end reliable computing that boasted you could write your
> app on an Ultra and move it to a SUNFIRE. IBM now has that message
> because now you can write your app on Linux and run it on a Mainframe.
> Fun too how much money IBM makes from JAVA.
Probably won't see Sun (anything) on the mainframe. But since the
mainframe now runs Linux and since Sun fully supports and encourages
the use of Linux now, all IBM is doing is providing another
fully Sun compatible platform to Sun.
Sunfire will continue to live on since there isn't a reasonable
"cheap" platform alternative. Again, the high end is where Sparc
will continue to live on for a while. As commodity x86 platforms
become available with hundreds of processors, I'm sure Sun will
change. But.. when you consider that high end SMP is somewhat
of a niche market nowadays, perhaps it doesn't matter too much
what happens to the Sunfire line.
If you had written this post a year ago... it would have
been somewhat valid. But as of today, Sun is a Linux company.
Somebody from Sun may feel free to provide more details
on Sun's current Linux strategy. In many ways, Sun is a much
bigger supporter of Linux than IBM. If you doubt that, I
encourage you to call Sun and ask them.
Don Davis
I think everyone is missing the point. Hardware and operating systems
are necessary evil's in providing an environment for an application to
run.
Applicatons are where businesses make their money. Until vendors stop
charging on a per processor basis, (real or virtual if using LPAR,
micro partitioning, VMWare... etc) the cost of the app is what
matters. In the meantime, from a cost standpoint, I want the world's
fastest one-way processor. If that's not fast enough to run the app,
the next best alternative is the world's fastest two-way.... etc.
It makes no sense to acquire a low cost hardware platform (choose your
vendor),
a free or low cost operating system, (choose your OS), then fork over
several hundred thousand dollars for an enterprise class database
product. (Other software products are also moving to number of CPU
based pricing, or charging by the client, if client/server)
What I would really like to see is an open source database (like
MySQL) with
advanced features, like triggers, stored procedures.. etc. This would
take most
of the price gouging out of software sales, and make hardware and OS
software important again. Right now, the harware/OS software component
is such a small fraction of the total costs of ownership, I'm not sure
it really matters. Both have reached commidity status. Just my
opinion, comments welcome.
Rich Teer
> Hey Rich - the grin looks remarkably as I remember you!
Excellent - all that cash spent of self-portrait art classes
wasn't wasted! :-)
Rich Teer
> What I would really like to see is an open source database (like
> MySQL) with
> advanced features, like triggers, stored procedures.. etc. This would
It's already here, and it's called PostgreSQL.
Richard D. Latham
< snip >
>
> Sunfire will continue to live on since there isn't a reasonable
> "cheap" platform alternative. Again, the high end is where Sparc
> will continue to live on for a while. As commodity x86 platforms
> become available with hundreds of processors, I'm sure Sun will
> change. But.. when you consider that high end SMP is somewhat
> of a niche market nowadays, perhaps it doesn't matter too much
> what happens to the Sunfire line.
... "high end SMP is somewhat of a niche market nowadays" ...
I wish I had enough money that I could consider a multi-billion dollar
a year market was a "niche market" :-)
--
#include <disclaimer.std> /* I don't speak for IBM ... */
/* Heck, I don't even speak for myself */
/* Don't believe me ? Ask my wife :-) */
Richard D. Latham lat...@us.ibm.com
Rich Teer
> Actually... I have word from the inside of Sun that Sun is "full tilt"
> favoring Linux right now. Solaris will be an additional offering,
Well, we obviously have different inside contacts, and I don't
recall hearing your voice on various NDA conference calls discussing
this matter I've been party to...
> but after it is completely open sourced, users will have a choice
> between FULLY Sun supported LINUX (e.g. JDS - which is built on
> Novell SUSE Linux Enterprise Server v8), or FULLY Sun supported
> Solaris.
JDS will soon be available on Solaris - both SPARC and x86. It is
NOT tied to a Linuc kernel.
> Your information is wrong. Though many in the newsgroup will likely
> support the idea of Sun NOT going after Linux, my inside sources
> say that Linux has become king at Sun currently.
I think your sources are dated. Yes, Linux is supported in and
by Sun, but they are refocussing their priorities on Solaris
(SPARC and x86).
> If you had written this post a year ago... it would have
> been somewhat valid. But as of today, Sun is a Linux company.
I disagree. Sun fully supports Linux, yes, but their focus
is on Solaris. I see that message becoming even louder and
clearer when S10 hits FCS...
CJT
> Chris Cox <ccox_nop...@airmail.net> writes:
>
> < snip >
>
>>Sunfire will continue to live on since there isn't a reasonable
>>"cheap" platform alternative. Again, the high end is where Sparc
>>will continue to live on for a while. As commodity x86 platforms
>>become available with hundreds of processors, I'm sure Sun will
>>change. But.. when you consider that high end SMP is somewhat
>>of a niche market nowadays, perhaps it doesn't matter too much
>>what happens to the Sunfire line.
>
>
> ... "high end SMP is somewhat of a niche market nowadays" ...
>
> I wish I had enough money that I could consider a multi-billion dollar
> a year market was a "niche market" :-)
>
I wish I had enough money to afford one of those high end SMP machines.
I'd happily admit to being a niche if that's all it took.
Chris Thompson
The snapshot of the author in Rich's forthcoming book should be
really interesting ... :-)
Chris Thompson
Email: cet1 [at] cam.ac.uk
Chris Cox
> On Mon, 26 Jul 2004, Chris Cox wrote:
>
>
>>Actually... I have word from the inside of Sun that Sun is "full tilt"
>>favoring Linux right now. Solaris will be an additional offering,
>
>
> Well, we obviously have different inside contacts, and I don't
> recall hearing your voice on various NDA conference calls discussing
> this matter I've been party to...
So... you're NDS info says that Sun says "Linux sucks??" Probably
not. Certainly not the message being communicated internally
at Sun and certainly not the message being communicated outside
of Sun.
Though I did expect you to respond to my comments... shame it
wasn't somebody officially from Sun though.
>
>
>>but after it is completely open sourced, users will have a choice
>>between FULLY Sun supported LINUX (e.g. JDS - which is built on
>>Novell SUSE Linux Enterprise Server v8), or FULLY Sun supported
>>Solaris.
>
>
> JDS will soon be available on Solaris - both SPARC and x86. It is
> NOT tied to a Linuc kernel.
Actually JDS is more tied to a Linux distribution (which is JDS).
JDS is a Linux distribution. Not some kind of wrapper on top
of *ix. Those are Sun's words...
>
>
>>Your information is wrong. Though many in the newsgroup will likely
>>support the idea of Sun NOT going after Linux, my inside sources
>>say that Linux has become king at Sun currently.
>
>
> I think your sources are dated. Yes, Linux is supported in and
> by Sun, but they are refocussing their priorities on Solaris
> (SPARC and x86).
This information is approximately 1-2 weeks old now. I suppose
that's a long time in technology.
>
>
>>If you had written this post a year ago... it would have
>>been somewhat valid. But as of today, Sun is a Linux company.
>
>
> I disagree. Sun fully supports Linux, yes, but their focus
> is on Solaris. I see that message becoming even louder and
> clearer when S10 hits FCS...
>
So we are at an impasse. Rich says Sun is the same ole Sun
focusing on Solaris. Sun executives (via Sun employees to me) say
otherwise. Who is right? Anyone else have better info than this?
The reason I posted like I did is that I made a post in another
forum basically saying EXACTLY what you've said here and I was
contacted directly by someone from Sun who set me straight
on the matter. I didn't want to believe them either.
Chris Cox
> Richard D. Latham wrote:
>
>> Chris Cox <ccox_nop...@airmail.net> writes:
>>
>> < snip >
>>
>>> Sunfire will continue to live on since there isn't a reasonable
>>> "cheap" platform alternative. Again, the high end is where Sparc
>>> will continue to live on for a while. As commodity x86 platforms
>>> become available with hundreds of processors, I'm sure Sun will
>>> change. But.. when you consider that high end SMP is somewhat
>>> of a niche market nowadays, perhaps it doesn't matter too much
>>> what happens to the Sunfire line.
>>
>>
>>
>> ... "high end SMP is somewhat of a niche market nowadays" ...
>>
>> I wish I had enough money that I could consider a multi-billion dollar
>> a year market was a "niche market" :-)
>>
> I wish I had enough money to afford one of those high end SMP machines.
> I'd happily admit to being a niche if that's all it took.
>
Hmmm... how many companies have an E10K, E15K?? While these units
do sell... it's still a small number of companies who actually
need this kind of high end powerhouse.
Similary, how many companies have a mainframe?? Same arguments.
Don't confuse "niche market" with an unprofitable one. IMHO,
always higher margins to be had with the niche markets. It's
not a bad business to be in.
Brian Utterback
It's two mints in one.
Sun is a very large company. There are a large number of people that are focused
on Linux as a product. There are a large number of people focused on Linux as
a platform among others. And there are a large number of people focused on
Solaris.
There is a lot going on in the Solaris front, what with all the innovations
in Solaris 10 and the open sourcing project. Of course there is more focus
on Solaris, since just about all the Solaris kernel developers are here. Linux
doesn't directly require so many resource directly, even if the plans are large.
--
blu
It is bafflingly paradoxical that the United States is by far the world's
leading scientific nation while simultaneously housing the most scientifically
illiterate populace outside the Third World. - Richard Dawkins
--------------------------------------------------------------------------------
Brian Utterback - OP/N1 Revenue Product Engineering, Sun Microsystems, Inc.
Ph/VM: 877-259-7345, Em:brian.utterback-at-ess-you-enn-dot-kom
UNIX admin
> clinging on to Solaris instead of switching to Linux OR having a Linux
> affinity Solaris such as something that could be branded "SolarisLX"
> that would work seemlessly with Linux. Make Linux or Linux affinity
> Solaris availible for everything from a SUNFIRE to the Ultrasparc.
Allow me to be blunt: you're trippin'.
Why would Sun dump a reliable, robust OS whose functionality Linux has yet
to achieve? That would not be prudent, to say the least. Some of Sun's
biggest customers are largest banks in the world, and when you mention Linux
to them, their hair stands up on their heads (first hand experience).
That's just one example. There are many.
Linux isn't even close in robustness and functionality of Solaris, and it'll
be "a while" before it gets there. But who knows how far ahead will Solaris
be then?
Rich Teer
> The snapshot of the author in Rich's forthcoming book should be
> really interesting ... :-)
Thankfully, the cover DOESN'T have my photo on it. I'm an ugly
bugger, and I didn't want to put off potential purchasers! :-)
My ugly mug will viewable by all and sundry when I upload my
new, improved web site, in the next few days. The new-look
site will also contain the book's official site, wherein all
the usual goodies (example source code, Preface, and sample
chapter) will be found.
To paraphrase Scott and Jonathan, this is not an announcement.
That will be in a few days, when there's actually something
to look at... :-)
Rich Teer
> So... you're NDS info says that Sun says "Linux sucks??" Probably
No, but there's a big gap between "Linux sucks" and "Linux rules,
and is preferred over Solaris".
> Though I did expect you to respond to my comments... shame it
> wasn't somebody officially from Sun though.
Heh - by the time an official response got vetted by SUn's
land sharks, we'd have moved onto something else! :-)
> Actually JDS is more tied to a Linux distribution (which is JDS).
> JDS is a Linux distribution. Not some kind of wrapper on top
> of *ix. Those are Sun's words...
I see. I must have mis-read my invitation to JDS on Solaris Beta,
then. Nope, it clearly says Solaris. Last I checked, Solaris was
not Linux (I think Linus would have a word or two to say if Sun
tried to claim that!).
> So we are at an impasse. Rich says Sun is the same ole Sun
> focusing on Solaris. Sun executives (via Sun employees to me) say
> otherwise. Who is right? Anyone else have better info than this?
I guess time will tell, either way...
Chris Cox
...
Rich,
Do you know of anything on the web that explains better the
position of Sun on this? If things are really as you
have said, I need to get some hard data to certain
folks around here (so as to not mislead them).
Sounds like JDS on Linux was a stop gap measure until
their true JDS on Solaris product could be developed (???).
I really need to know this. Don't want folks to go
down the Linux trail needlessly. If the ultimate goal
is JDS on Solaris... I want to know for sure.
Chris Cox
>
...
> It's two mints in one.
>
> Sun is a very large company. There are a large number of people that are
> focused
> on Linux as a product. There are a large number of people focused on
> Linux as
> a platform among others. And there are a large number of people focused on
> Solaris.
I hear what you are saying, and perhaps Sun is actively trying to
pursuade folks to both strategies. I've seen stranger things.
>
> There is a lot going on in the Solaris front, what with all the innovations
> in Solaris 10 and the open sourcing project. Of course there is more focus
> on Solaris, since just about all the Solaris kernel developers are here.
> Linux
> doesn't directly require so many resource directly, even if the plans
> are large.
>
Hmmm.. my problem is trying to figure out what the TRUE direction
is. Even is Sun is seemingly 100% Linux and 100% Solaris, eventually,
one will out favor the other. What I think I hear you saying is
that Solaris is favored.... but that's certainly not what
Sun's marketing and sales folks are saying.... well at least
not from what I've been told.
I haven not been told that Sun isn't Solaris anymore... but
certainly I have been told that Sun is totally behind their
JDS LINUX (emphasis on LINUX) product. If JDS is migrating
to a Solaris based version, this could make a huge difference
with regards to software and hardware choices being made.
Not sure if Sun understands that.
Frank Cusack
> On Mon, 26 Jul 2004, Chris Cox wrote:
>> Actually JDS is more tied to a Linux distribution (which is JDS).
>> JDS is a Linux distribution. Not some kind of wrapper on top
>> of *ix. Those are Sun's words...
>
> I see. I must have mis-read my invitation to JDS on Solaris Beta,
> then. Nope, it clearly says Solaris. Last I checked, Solaris was
> not Linux (I think Linus would have a word or two to say if Sun
> tried to claim that!).
Indeed, JDS has nothing (zilch, zero, nada) to do with Linux other
than that happens to be an OS it runs *on top of*.
I think it's really funny that you folks are arguing about whether or
not JDS is a Linux distro, with "Linux" not being part of the name,
whereas the more interesting (in the laffable and dare I say
embarrasing Sun marketing way) conflict is that it has nothing to do
with "Java" which is the "J"!
Especially ironic because I'm sure Sun would sue anyone else trying
to distribute a non-Java product that carried Java in the title. Didn't
Sun already sue MicroSoft for just such a thing?
Let's rebrand FreeBSD as javaOS and just make sure tomcat is part of
the base install. That'd be pretty close to the JDS in terms of name
misuse.
/fc
Benjamin Gawert
> Hmmm... how many companies have an E10K, E15K??
You'd be surprised how many E15ks (E10k is old, slow and outdated) are
around, or IBM p690's, or HP Superdomes...
> While these units
> do sell... it's still a small number of companies who actually
> need this kind of high end powerhouse.
Not that small as You might think...
> Similary, how many companies have a mainframe?? Same arguments.
Well, from what I know IBM still makes most of its money from Mainframes...
Benjamin
Chris Cox
...
>>Similary, how many companies have a mainframe?? Same arguments.
>
>
> Well, from what I know IBM still makes most of its money from Mainframes...
>
mean low margins even if we're talking low volumes.
Even though the last comment was somewhat IBM related, I'm adjusting
followups to just the solaris and linux groups.
Chris Cox
> On Mon, 26 Jul 2004 19:36:46 GMT Rich Teer <rich...@rite-group.com> wrote:
>
>>On Mon, 26 Jul 2004, Chris Cox wrote:
>>
>>>Actually JDS is more tied to a Linux distribution (which is JDS).
>>>JDS is a Linux distribution. Not some kind of wrapper on top
>>>of *ix. Those are Sun's words...
>>
>>I see. I must have mis-read my invitation to JDS on Solaris Beta,
>>then. Nope, it clearly says Solaris. Last I checked, Solaris was
>>not Linux (I think Linus would have a word or two to say if Sun
>>tried to claim that!).
>
>
> Indeed, JDS has nothing (zilch, zero, nada) to do with Linux other
> than that happens to be an OS it runs *on top of*.
>
> I think it's really funny that you folks are arguing about whether or
> not JDS is a Linux distro, with "Linux" not being part of the name,
> whereas the more interesting (in the laffable and dare I say
> embarrasing Sun marketing way) conflict is that it has nothing to do
> with "Java" which is the "J"!
Now that is funny! Unfortunately true though (for the most part).
>
> Especially ironic because I'm sure Sun would sue anyone else trying
> to distribute a non-Java product that carried Java in the title. Didn't
> Sun already sue MicroSoft for just such a thing?
>
> Let's rebrand FreeBSD as javaOS and just make sure tomcat is part of
> the base install. That'd be pretty close to the JDS in terms of name
> misuse.
>
Redirected followup to the solaris and linux groups. Let's not
abuse the bsd folks.
Richard L. Hamilton
aix...@yahoo.com (Don Davis) writes:
[...]
> What I would really like to see is an open source database (like
> MySQL) with
> advanced features, like triggers, stored procedures.. etc. This would
http://firebird.sourceforge.net/
http://www.ibphoenix.com/main.nfs?a=ibphoenix&s=1090876179:23108&page=vul_development
(the latter a major rewrite by the original developer to provide 64-bit SMP
support)
etc.
Rich Teer
> Rich,
> Do you know of anything on the web that explains better the
> position of Sun on this? If things are really as you
> have said, I need to get some hard data to certain
> folks around here (so as to not mislead them).
The best I can do using easy-to-find publically available
info is this (from http://wwws.sun.com/software/javadesktopsystem/faq.html ):
1. Q. What is the Sun Java Desktop System?
A. [...]
Future releases of Java Desktop System are planned to
support workstations and Sun Ray thin clients running
the Solaris Operating System.
and more definately:
4. Q. What is the Future Product Roadmap?
A. Q3CY2004:
Java Desktop System, Release 2 for the Solaris x86
platform for Sun Opteron-based Workstations only
Q4CY2004:
Java Desktop System, Release 3 introduces the following features:
* Support for the Solaris Operating System
[...]
I wouldn't be surprised if the release of JDSv3 coincided
with S10 FCS...
> Sounds like JDS on Linux was a stop gap measure until
> their true JDS on Solaris product could be developed (???).
Not quite. JDS will eventually run on top of Linux as well as
Solaris, in much the same way that StarOffice does today. I
guess Linux was the easier target (presumably, many components
have various Linuxisms that make a straight port a bit problematic),
which is why it got released first.
> I really need to know this. Don't want folks to go
> down the Linux trail needlessly. If the ultimate goal
> is JDS on Solaris... I want to know for sure.
I think the ultimate goal is to support JDS equally on Linux
and Solaris (both SPARC and x86), the customer making the
final choice.
The Ghost In The Machine
<mikeco...@yahoo.com>
wrote
on 25 Jul 2004 18:01:22 -0700
<3d6111f1.04072...@posting.google.com>:
[snip]
Erm, don't you mean Intel or Microsoft? (And even Intel
is suspect; Intel chips may run Microsoft Windows (and
be the only chips that run Microsoft Windows), but Intel
doesn't like Microsoft, IIRC.)
--
#191, ewi...@earthlink.net
It's still legal to go .sigless.
GreyCloud
Chris Cox wrote:
The government buys them and they aren't profitable. :-))
> IMHO,
> always higher margins to be had with the niche markets. It's
> not a bad business to be in.
>
--
---------------------------------
My other computer is a VAX.
CJT
> CJT wrote:
>
>> Richard D. Latham wrote:
>>
>>> Chris Cox <ccox_nop...@airmail.net> writes:
>>>
>>> < snip >
>>>
>>>> Sunfire will continue to live on since there isn't a reasonable
>>>> "cheap" platform alternative. Again, the high end is where Sparc
>>>> will continue to live on for a while. As commodity x86 platforms
>>>> become available with hundreds of processors, I'm sure Sun will
>>>> change. But.. when you consider that high end SMP is somewhat
>>>> of a niche market nowadays, perhaps it doesn't matter too much
>>>> what happens to the Sunfire line.
>>>
>>>
>>>
>>>
>>> ... "high end SMP is somewhat of a niche market nowadays" ...
>>>
>>> I wish I had enough money that I could consider a multi-billion dollar
>>> a year market was a "niche market" :-)
>>>
>> I wish I had enough money to afford one of those high end SMP machines.
>> I'd happily admit to being a niche if that's all it took.
>>
>
> Hmmm... how many companies have an E10K, E15K?? While these units
> do sell... it's still a small number of companies who actually
> need this kind of high end powerhouse.
Thousands?
>
> Similary, how many companies have a mainframe?? Same arguments.
>
Tens of thousands?
> Don't confuse "niche market" with an unprofitable one. IMHO,
> always higher margins to be had with the niche markets. It's
> not a bad business to be in.
>
ro...@luther.kurtwerks.com
> Mike Cox wrote:
[deletia]
> become available with hundreds of processors, I'm sure Sun will
> change. But.. when you consider that high end SMP is somewhat
> of a niche market nowadays, perhaps it doesn't matter too much
> what happens to the Sunfire line.
Er, that's one heck of a big "niche," then.
Kurt
--
Government lies, and newspapers lie, but in a democracy they are
different lies.
ro...@luther.kurtwerks.com
[high-end SMP == "niche" market]
> Hmmm... how many companies have an E10K, E15K?? While these units
> do sell... it's still a small number of companies who actually
> need this kind of high end powerhouse.
>
> Similary, how many companies have a mainframe?? Same arguments.
Boatloads. I've been in a lot of datacenters and machine rooms since
1993. *Every* one of them had a least one piece of traditional big
iron or one or more "high-end SMP UNIX systems, where "high-end"
means >8 CPUs. Most of these datacenters and machine rooms had
multiple mainframes, multiple SMP UNIX systems, or both. One even
had four SP/2s, in addition to assorted traditional SMP systems
and the odd mainframes. The point being, of course, that it's more
common than you might think.
> Don't confuse "niche market" with an unprofitable one. IMHO,
> always higher margins to be had with the niche markets. It's
> not a bad business to be in.
Indeed.
Alan Coopersmith
|Actually... I have word from the inside of Sun that Sun is "full tilt"
|favoring Linux right now. Solaris will be an additional offering,
|between FULLY Sun supported LINUX (e.g. JDS - which is built on
|Novell SUSE Linux Enterprise Server v8), or FULLY Sun supported
|Solaris.
|
|support the idea of Sun NOT going after Linux, my inside sources
|say that Linux has become king at Sun currently.
Your inside sources must not know most of the people I work with or for
(admittedly in the Solaris division), nor do they listen to the same
executive comments that caused this article to be written:
http://www.vnunet.com/news/1156903
--
________________________________________________________________________
Alan Coopersmith * al...@alum.calberkeley.org * Alan.Coo...@Sun.COM
http://www.csua.berkeley.edu/~alanc/ * http://blogs.sun.com/alanc/
Working for, but definitely not speaking for, Sun Microsystems, Inc.
Gavin Maltby
> But you are right in Sun & solaris getting squeezed. Linux & bsd lack some
> functionality of solaris but cost near $0 . Its what happens when operating
> systems become commodities.
My once-yearly excursion into advocacy arguments. Slow learner.
I'm sure this has been discussed countless times before. Nonetheless,
I foolishly hit a few dictionary sites ...
Commodity:
1. A generic, mostly unprocessed, good that can be processed and resold
2. A useful item that can be used to commercial or other advantage
3. Something affording convenience, advantage or profit
4. advantage, or benefit
5. a class of economic goods
Now I'd suggest that the quote above is using the first form, suggesting that
operating systems have become generic - differing little from version to
version, offering much the same set of functionality, same quality, same
this that and everything. I'd agree that if OS were commodities in this narrow
sense that there'd be little room for Solaris or any other OS to
distinguish itself from others.
But I can't buy in to that narrowminded definition. I'd say that OS have
become commodities in the sense that they're a class of economic goods,
lots of people have one, they all have a similar set of core features to
create an application runtime environment etc. But there is still plenty
of room for operating systems to distinguish themselves: quality, testing,
reputation, stability, features, binary compatability across releases/patches,
ease of maintenance, ease of installation/configuration, tuning,
debugging, facilitation of development environment, availability
features, scalability, price, performance, licensing model,
open source, hardware support, available applications, ease of
administration, support, etc etc (I'm sure I've left out some key ones,
the point is there are many areas).
Now you can take any OS and hold it against those categories, and I guess
anyone so doing should also weight the categories in advance according
to what is important to them. Across the category scores there is no
one OS today that is a clear leader in all; even when weighted I'm
certain there is no one winner for all situations. Thinking just of
Solaris for a moment I can see some (I'd say many) areas where it's the
leader or right up there, and others where it's less good or perhaps not
at all good. Whether it is the correct choice for a particular
application depends on whether its stronger features align with
your requirements ratings, and it's weaker points are only a detractor
of those aspects are important to you. It's up to the OS vendor to
realize their strengths and appropriate markets, and their weaknesses
and corresponding missed market opportunities.
Another underlying assertion in quotes such as the above would also
seem to be that lists of features/categories etc as above are
either achieved or not achieved by a particular OS - there are no stages
in between, and suggesting there's no way beyond the state-of-the-art.
It's almost as if the underlying belief is that "how to write an OS"
or "how to introduce feature X to an OS" used to be well kept
commercial secrets with the accompanying assumption that these things
were really really difficult/complex, yet now these secrets have been
divulged and the real truth - that it's dead easy and everyone can
do it - has been exposed. The truth is that it's generally not
trivial to do well, that it is complex, that maturity and evolution
really do count, that feature quality really isn't a boolean quantity,
not everybody can do it (or if they do they don't do it even nearly
equally well) etc.
Now I hope that anyone who has bothered to read this far has noticed that
I've not chosen to explicitly sing the praises or highlight the
deficiencies of any one OS distribution. They all have their
strengths and weaknesses. My beef is against the implication that
OS being a "commodity" even begins to suggest that all OS are
approximately equal or can be easily made so. But to come off the
fence a little, I do very firmly believe that Solaris 10 (or Solaris Express
for the time being) is a very strong performer in many of the categories and
makes itself a particularly worthy contender in many (but not all) markets.
But I do recognise that there are areas where others lead the way or
are on a par - we won't be staying still in all those areas, either.
Gavin
Speaking for myself and not necessarily for Sun
Greg Copeland
> In article <f45a2235.0407...@posting.google.com>,
> aix...@yahoo.com (Don Davis) writes:
> [...]
>> What I would really like to see is an open source database (like
>> MySQL) with
>> advanced features, like triggers, stored procedures.. etc. This would
> [...]
>
A native windows port should be out in the next couple of months. Alphas
of the port have been available for some time. I believe they are now
officially considered to be betas at this point.
Just FYI, from the SP perspective, with PostgreSQL, you can write stored
procedures in Perl, Python, TCL, C/C++, Java (I think, I know they've been
working on it), and it's own native, Oracle-like, langauge. I'm sure I'm
missing several language options there too.
Cheers,
Greg
JEDIDIAH
On 2004-07-26, Richard D. Latham <lat...@us.ibm.com> wrote:
> Chris Cox <ccox_nop...@airmail.net> writes:
>
>< snip >
>>
>> Sunfire will continue to live on since there isn't a reasonable
>> "cheap" platform alternative. Again, the high end is where Sparc
>> will continue to live on for a while. As commodity x86 platforms
>> change. But.. when you consider that high end SMP is somewhat
>> of a niche market nowadays, perhaps it doesn't matter too much
>> what happens to the Sunfire line.
>
>
> I wish I had enough money that I could consider a multi-billion dollar
> a year market was a "niche market" :-)
That's what? 1000 machines?
That's pretty much a niche.
--
It is not true that Microsoft doesn't innovate.
They brought us the email virus.
In my Atari days, such a notion would have |||
been considered a complete absurdity. / | \
JEDIDIAH
> In comp.unix.solaris Chris Cox <ccox_nop...@airmail.net> wrote:
>> CJT wrote:
>
> [high-end SMP == "niche" market]
>
>> Hmmm... how many companies have an E10K, E15K?? While these units
>> do sell... it's still a small number of companies who actually
>> need this kind of high end powerhouse.
>>
>> Similary, how many companies have a mainframe?? Same arguments.
>
> Boatloads. I've been in a lot of datacenters and machine rooms since
> 1993. *Every* one of them had a least one piece of traditional big
> iron or one or more "high-end SMP UNIX systems, where "high-end"
> means >8 CPUs. Most of these datacenters and machine rooms had
> multiple mainframes, multiple SMP UNIX systems, or both. One even
> had four SP/2s, in addition to assorted traditional SMP systems
> and the odd mainframes. The point being, of course, that it's more
> common than you might think.
It's probably getting less common. As the big application vendors
(like Oracle) for machines such as those push clustering more and
more, such beasts may go the way of the dinosaurs.
NUMA boxes are much more expensive per MIP than their smaller cousins.
This expense is enough to motivate change.
Sure, it's a lucrative niche. Therein lies the problem.
JEDIDIAH
> mikeco...@yahoo.com (Mike Cox) wrote in message news:<3d6111f1.04072...@posting.google.com>...
[deletia]
>> SUN was high-end reliable computing that boasted you could write your
>> app on an Ultra and move it to a SUNFIRE. IBM now has that message
>> because now you can write your app on Linux and run it on a Mainframe.
>> Fun too how much money IBM makes from JAVA.
>
> I think everyone is missing the point. Hardware and operating systems
> are necessary evil's in providing an environment for an application to
> run.
>
> Applicatons are where businesses make their money. Until vendors stop
> charging on a per processor basis, (real or virtual if using LPAR,
> micro partitioning, VMWare... etc) the cost of the app is what
> matters. In the meantime, from a cost standpoint, I want the world's
> fastest one-way processor. If that's not fast enough to run the app,
> the next best alternative is the world's fastest two-way.... etc.
>
> It makes no sense to acquire a low cost hardware platform (choose your
> vendor),
> a free or low cost operating system, (choose your OS), then fork over
> several hundred thousand dollars for an enterprise class database
> product. (Other software products are also moving to number of CPU
> based pricing, or charging by the client, if client/server)
Sure it does. The low cost operating system may be optimized by the
high cost application vendor to operate best with the application with
minimial extra cost.
This is where Free Beer and Free-as-in-speech converge.
Oracle can contribute to the Linux kernel in an effort to get more
of shrinking IT budgets for themselves. They can push out Sun and
they can push out Veritas giving their customers an apparent bargain
while still getting their usual arm+leg.
>
> What I would really like to see is an open source database (like
> MySQL) with
> advanced features, like triggers, stored procedures.. etc. This would
> take most
> of the price gouging out of software sales, and make hardware and OS
If you think triggers and stored procedures are what make an enterprise
RDBMS advanced then you have no real understanding of the problem domain.
> software important again. Right now, the harware/OS software component
> is such a small fraction of the total costs of ownership, I'm not sure
> it really matters. Both have reached commidity status. Just my
> opinion, comments welcome.
Robert Mazur
> What I would really like to see is an open source database (like
> MySQL) with
> advanced features, like triggers, stored procedures.. etc.
Off topic, but for the record MySQL's version 5, currently in alpha, has
stored procedures and *I believe* triggers.
Rob
Chris Cox
...
> I think the ultimate goal is to support JDS equally on Linux
> and Solaris (both SPARC and x86), the customer making the
> final choice.
>
Thanks Rich... but I think you know that Sun will favor
one or the other eventually. Can't imagine them leaving
Solaris and Linux on equal footing. Just my opinion.
Erik Magnuson
>
> > Sounds like JDS on Linux was a stop gap measure until
> > their true JDS on Solaris product could be developed (???).
>
> Not quite. JDS will eventually run on top of Linux as well as
> Solaris, in much the same way that StarOffice does today. I
> guess Linux was the easier target (presumably, many components
> have various Linuxisms that make a straight port a bit problematic),
> which is why it got released first.
I would hazard a guess that Linux was supported first due to a more
diverse set of drivers available for Linux - basing JDS on Solaris x86
would require some major scrambling on Sun's part to broaden the HCL for
Solaris. I wouldn't be surprised if Linuxisms were also an issue.
I'm hoping that one fall-out of porting Solaris to AMD64 is a broader
range of drivers for Solaris Sparc whether written by Sun or the device
manufacturer.
-Erik
Joerg Schilling
Alan Coopersmith <al...@alum.calberkeley.org> wrote:
>|Your information is wrong. Though many in the newsgroup will likely
>|support the idea of Sun NOT going after Linux, my inside sources
>|say that Linux has become king at Sun currently.
>
>Your inside sources must not know most of the people I work with or for
>(admittedly in the Solaris division), nor do they listen to the same
>executive comments that caused this article to be written:
> http://www.vnunet.com/news/1156903
From my contacts to Sun, I also do not see any bit Linux hype....
It may be that some of the Sun marketing people write too much about Linux.
--
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
Brian Utterback
The goal is JDS on the desktop. Not Linux on the desktop, not Solaris on
the desktop, JDS on the desktop. Sun doesn't want a Solaris shop or a
Linux shop; we want a Sun shop!
--
blu
It is bafflingly paradoxical that the United States is by far the world's
leading scientific nation while simultaneously housing the most scientifically
illiterate populace outside the Third World. - Richard Dawkins
--------------------------------------------------------------------------------
Brian Utterback - OP/N1 Revenue Product Engineering, Sun Microsystems, Inc.
Ph/VM: 877-259-7345, Em:brian.utterback-at-ess-you-enn-dot-kom
Chris Cox
...snip...
> The goal is JDS on the desktop. Not Linux on the desktop, not Solaris on
> the desktop, JDS on the desktop. Sun doesn't want a Solaris shop or a
> Linux shop; we want a Sun shop!
Does this imply more of a shift away from hardware and OS's for
Sun? Will JES be on Linux as well as Solaris? Just curious.
Rich Teer
> Sun? Will JES be on Linux as well as Solaris? Just curious.
I believe it already is (HP-SUX too, IIRC).
GreyCloud
Rich Teer wrote:
> On Wed, 28 Jul 2004, Chris Cox wrote:
>
>
>>Sun? Will JES be on Linux as well as Solaris? Just curious.
>
>
> I believe it already is (HP-SUX too, IIRC).
>
What do you think of the new Sun Opteron box that was just released?
Priced around $2k.
Alan Hargreaves - Product Technical Support (APAC)
> On Mon, 26 Jul 2004, Don Davis wrote:
>
>
>>What I would really like to see is an open source database (like
>>MySQL) with
>
>
> It's already here, and it's called PostgreSQL.
>
Have a read of Alan Burlinson's experiences with databases at
http://blogs.sun.com/roller/comments/alanbur/Weblog/elbonian_database_technology
and the comments to his blog. I found it informative.
alan.
--
Alan Hargreaves - http://blogs.sun.com/tpenta
Senior Technical Support Specialist/VOSJEC Engineer
Product Technical Support (APAC)
Sun Microsystems
Rich Teer
> What do you think of the new Sun Opteron box that was just released?
> Priced around $2k.
I've not seen one in the flesh, but they do look VERY interesting,
the V2100z more so.
Perhaps someone from Sun will be kind enough to send me one,
so that I can review it? ;-)
GreyCloud
Rich Teer wrote:
> On Wed, 28 Jul 2004, GreyCloud wrote:
>
>
>>What do you think of the new Sun Opteron box that was just released?
>>Priced around $2k.
>
>
> I've not seen one in the flesh, but they do look VERY interesting,
> the V2100z more so.
>
> Perhaps someone from Sun will be kind enough to send me one,
> so that I can review it? ;-)
>
Of course. After you review it then you can add me to their discount
list. Is there any truth to the rumours that Sun is porting Solaris to
the Itanium2 chip?
Rich Teer
> Of course. After you review it then you can add me to their discount
How would I do that - I don't work for Sun! Suitably lucretive
offers are welcome, though... :-)
> list. Is there any truth to the rumours that Sun is porting Solaris to
> the Itanium2 chip?
I haven't the faintest idea. But considering that Sun had Solaris
running on an Itanic simulation a few years ago, I guess it wouldn't
be that hard, especially if it (Itanic2) is very similar to Opteron.
And if ISN'T similar to Opteron, I'd question the thinking behind
supporting Itanic2, given it's miniscule market size. The only
possible reason I can think of would be so that Sun can say they have
a migration path for HP users if/when HP-UX is EOLed.
Joerg Schilling
Rich Teer <rich...@rite-group.com> wrote:
>> list. Is there any truth to the rumours that Sun is porting Solaris to
>> the Itanium2 chip?
>
>I haven't the faintest idea. But considering that Sun had Solaris
>running on an Itanic simulation a few years ago, I guess it wouldn't
>be that hard, especially if it (Itanic2) is very similar to Opteron.
I would guess that Itanic2 is more or less Itanic compatible.
While 64 bit graphic card drivers, ... could be reused to 100%,
trap handling, MMU support, the linker (ld.so), mdb, kmdb, ....
need to be completely different.
Gavin Maltby
> Of course. After you review it then you can add me to their discount
> list. Is there any truth to the rumours that Sun is porting Solaris to
> the Itanium2 chip?
Seems to be in the news:
Has a quote from JS saying Sun is looking at this (so it must be true).
Gavin
John D Groenveld
Erik Magnuson <er...@tfb.com> wrote:
>I would hazard a guess that Linux was supported first due to a more
>diverse set of drivers available for Linux - basing JDS on Solaris x86
Hazardous guess.
If you Google Peder Ulander you'll find a better guess why MadHatter for
Solaris on the SunRay and Blade 2000 was an after-thought for the
Cobalt Networks alumni.
John
groe...@acm.org
Erik Magnuson
I was referring to why using Linux was preferable to using Solaris x86
for the initial OS for JDS. It seems obvious that Sun would have more
luck winning converts to JDS if there was a way of using existing
machines - especially as an escape from MS's new license plan.
Knowing a little about the history of Cobalt, it would be natural to
port to Linux first. Sun's buying Cobalt was not as smart a move as
when they bought FPS/Cray from SGI.
Ultimately, it would make sense to use a thin client and server
combination when the existing PC hardware wears out.
- Erik
GreyCloud
Gavin Maltby wrote:
Hmmm... the Itanium line won't be cheap tho.
GreyCloud
Rich Teer wrote:
> On Thu, 29 Jul 2004, GreyCloud wrote:
>
>
>>Of course. After you review it then you can add me to their discount
>
>
> How would I do that - I don't work for Sun! Suitably lucretive
> offers are welcome, though... :-)
>
>
>>list. Is there any truth to the rumours that Sun is porting Solaris to
>>the Itanium2 chip?
>
>
> I haven't the faintest idea. But considering that Sun had Solaris
> running on an Itanic simulation a few years ago, I guess it wouldn't
> be that hard, especially if it (Itanic2) is very similar to Opteron.
> And if ISN'T similar to Opteron, I'd question the thinking behind
> supporting Itanic2, given it's miniscule market size. The only
> possible reason I can think of would be so that Sun can say they have
> a migration path for HP users if/when HP-UX is EOLed.
>
From what I've read over in cov, HP-UX is pushing out Tru64-UNIX for
some reason or other. They are porting OpenVMS to Itanium2 and it seems
to be in testing for now, but the reports coming back in aren't all that
good. It seems that the port shows some areas of performance is pretty
bad, but I don't know which area. Some are speculating that the EV line
of Alphas may have to be restarted again if this port doesn't work well.
This is HPs chip design from the start so its hard to say what they'll do.
Paul Eggert
> Have a read of Alan Burlinson's experiences with databases at
> http://blogs.sun.com/roller/comments/alanbur/Weblog/elbonian_database_technology
Here's what I got, just now. The "elbonian_database_technology"
dig is cute, but it'd be cuter if the blog itself worked....
Internal Server Error
The server encountered an internal error or misconfiguration and was
unable to complete your request.
Please contact the server administrator, [name removed]@sun.com and
inform them of the time the error occurred, and anything you might
have done that may have caused the error.
More information about this error may be available in the server error log.
Apache/2.0.49 (Unix) mod_jk2/2.0.4 Server at blogs.sun.com Port 80
Andrew Gabriel
GreyCloud <mi...@cumulus.com> writes:
>
> From what I've read over in cov, HP-UX is pushing out Tru64-UNIX for
> some reason or other.
HP have ended up with lots of OS's, but mostly with far too little
market share to be viable. HP-UX is the one with the largest market
share of the bunch, and the theory was that all the others would
just move across making HP-UX more viable and reducing HP's support
costs. HP overlooked the fact that none of their non-HP-UX customers
would dream of moving to HP-UX, although this seemed obvious to
everyone else.
> They are porting OpenVMS to Itanium2 and it seems
> to be in testing for now, but the reports coming back in aren't all that
> good. It seems that the port shows some areas of performance is pretty
> bad, but I don't know which area. Some are speculating that the EV line
> of Alphas may have to be restarted again if this port doesn't work well.
> This is HPs chip design from the start so its hard to say what they'll do.
They are in the same mess with chips as they are with OS's, actually
probably even slightly worse. Volumes of HP-PA and Alpha were too low
to sustain development, so the idea was to switch to a cheap comodity
high volume processor, Itanium. The trouble is itanium has failed to
become a cheap comodity high volume processor -- actually, just about
everyone else other than HP has abandoned it (certainly everyone who
might have made it cheap comodity high volume). So, take HP's original
processor sales of Alpha and HP-PA which were too low to sustain the
chip development, and spead them across more processor families,
Alpha, HP-PA, Itanium, and you can see they're really in a mess with
even lower volumes on each processor making them all less viable.
Not to mention they are now pushing Opteron too, which although safe
in itself, further dilutes the viability of their low volume processors.
Now isn't a good time for HP customers, most of whom are being pushed
from one non-viable processor to another non-viable processor -- HP
has gone to a lot of effort to drive them up a No-Through-Road.
I don't think there's any chance of Alpha being restarted. The last
processor in the range was just released EV7z and that's it. HP no
longer has the Alpha development team anyway.
--
Andrew Gabriel
Consultant Software Engineer
Frank Cusack
> At Thu, 29 Jul 2004 17:07:03 +1000, "Alan Hargreaves - Product Technical Support (APAC)" <Alan.Ha...@Sun.COM> writes:
>
>> Have a read of Alan Burlinson's experiences with databases at
>> http://blogs.sun.com/roller/comments/alanbur/Weblog/elbonian_database_technology
>
> Here's what I got, just now. The "elbonian_database_technology"
> dig is cute, but it'd be cuter if the blog itself worked....
>
> Internal Server Error
Worked for me, earlier today. The summary is, Oracle is great, all
open source db's are missing features one takes for granted in a "real"
database.
Like Solaris vs. Linux.
/fc
GreyCloud
Andrew Gabriel wrote:
That makes sense. It seems that the way things are going that the only
processor left will be either the Intel family or IBM family. Suns
Sparc will be about the only one on its own.
> I don't think there's any chance of Alpha being restarted. The last
> processor in the range was just released EV7z and that's it. HP no
> longer has the Alpha development team anyway.
>
Did the Alpha development disappear?
Or did the people migrate to other companies?
John D Groenveld
Erik Magnuson <er...@tfb.com> wrote:
>Ultimately, it would make sense to use a thin client and server
>combination when the existing PC hardware wears out.
Ultimately? If Sun was selling to a CTO ripping out her call center
of a thousand 5 year old ThinkCenters running OS/2, allowing the
sales rep to sell the JDS on Sun Ray solution makes more cents.
Oh wait, the reason the MadHatters were porting Sun Ray Server to
Linux was to support Enterpise Linux Client.
John
groe...@acm.org
Andrew Gabriel
GreyCloud <mi...@cumulus.com> writes:
> Andrew Gabriel wrote:
>> I don't think there's any chance of Alpha being restarted. The last
>> processor in the range was just released EV7z and that's it. HP no
>> longer has the Alpha development team anyway.
>>
> Did the Alpha development disappear?
> Or did the people migrate to other companies?
IIRC, HP gave the Alpha development team to Intel as part of their
contribution to the Itanium effort.
--
Andrew Gabriel
Thomas Dehn
Yep, but seeing whats coming, quite a few of the Alpha developers
had left HP earlier.
Thomas
Benjamin Gawert
> I haven't the faintest idea. But considering that Sun had Solaris
> running on an Itanic simulation a few years ago, I guess it wouldn't
> be that hard, especially if it (Itanic2) is very similar to Opteron.
Itanium is totally different from the Opteron.
> And if ISN'T similar to Opteron, I'd question the thinking behind
> supporting Itanic2, given it's miniscule market size. The only
> possible reason I can think of would be so that Sun can say they have
> a migration path for HP users if/when HP-UX is EOLed.
Or that Sun users have a migration path with their current hardware if/when
Solaris would get EOL'd ;-)
Benjamin
Benjamin Gawert
> The trouble is itanium has failed to
> become a cheap comodity high volume processor -- actually, just about
> everyone else other than HP has abandoned it
Definitely no. There are more IA64 vendors than HP. Companies like SGI, NEC,
etc...
> Not to mention they are now pushing Opteron too, which although safe
> in itself, further dilutes the viability of their low volume
> processors.
Certainly not. HPs Opteron targets a much different market than the
PA-RISC/IA64 systems...
> Now isn't a good time for HP customers, most of whom are
> being pushed from one non-viable processor to another non-viable
> processor -- HP
> has gone to a lot of effort to drive them up a No-Through-Road.
Not really. The transition from PA-RISC to IA64 was not trouble-free but
quite smooth after all. IA64 is the way to go for HP. Even if Itanium
doesn't get a high volume chip (which isn't really HPs or intels intention)
it is still way cheaper than the older PA-RISC development. And the new
PA-8800 uses the same system components (i.e. chipsets) like Itanium2.
Opteron is in no way considered as a replacement for IA64 but placed below
the IA64 line.
Benjamin
GreyCloud
Benjamin Gawert wrote:
This is a new viewpoint to me. How much different is the performance
between the Itanium2 and the amd64 at a corporate level? Or is this
question not really answerable yet?
GreyCloud
Thomas Dehn wrote:
This news really isn't what I was expecting. Actually, I find it a bit
disturbing that a shakeout of some kind is taking place and we are left
with one or two processor chip makers left.
Andrew Gabriel
"Benjamin Gawert" <bga...@gmx.de> writes:
> Andrew Gabriel wrote:
>
>> The trouble is itanium has failed to
>> become a cheap comodity high volume processor -- actually, just about
>> everyone else other than HP has abandoned it
>
> Definitely no. There are more IA64 vendors than HP. Companies like SGI, NEC,
> etc...
Which is what I said in the very next bit you snipped.
However, their volumes are totally insignificant, even all added
togther.
>> Not to mention they are now pushing Opteron too, which although safe
>> in itself, further dilutes the viability of their low volume
>> processors.
>
> Certainly not. HPs Opteron targets a much different market than the
> PA-RISC/IA64 systems...
Sure -- Opteron has very successfully taken the volume market IA64 was
aiming at. That's why Intel had to do a big U-turn and bring Yamhill/EM64T
into production, to avoid missing the volume 64 bit market completely.
>> Now isn't a good time for HP customers, most of whom are
>> being pushed from one non-viable processor to another non-viable
>> processor -- HP
>> has gone to a lot of effort to drive them up a No-Through-Road.
>
> Not really. The transition from PA-RISC to IA64 was not trouble-free but
> quite smooth after all. IA64 is the way to go for HP. Even if Itanium
> doesn't get a high volume chip (which isn't really HPs or intels intention)
Oh it most certainly was, but they now recognise that's not going to
happen! If Itanium doesn't become a high volume chip, there's no money
to pay for future development, and the chip dies when Intel decides to
stop subsidising it. The volumes which HP buys won't come close to
making it a volume chip -- after all, that was why HP-PA was not viable.
> it is still way cheaper than the older PA-RISC development. And the new
Itanium is the most expensive CPU ever developed, and by a long way.
> PA-8800 uses the same system components (i.e. chipsets) like Itanium2.
>
> Opteron is in no way considered as a replacement for IA64 but placed below
> the IA64 line.
Opteron already took almost all IA64's marketplace.
Paul Eggert
>>> http://blogs.sun.com/roller/comments/alanbur/Weblog/elbonian_database_technology
> The summary is, Oracle is great, all open source db's are missing
> features one takes for granted in a "real" database.
Well, more accurately, Postgres itself worked fine for his example but
the PerlDBI interface to Postgres was lacking.
Personally, I've found that "Oracle is great" is true only if you
already have an Oracle wizard around to make your installation act
great on your problem. We've had our fair share of problems porting
applications to Oracle. If you're used to Oracle, it's great, yes.
Benjamin Gawert
> This is a new viewpoint to me. How much different is the performance
> between the Itanium2 and the amd64 at a corporate level? Or is this
> question not really answerable yet?
It depends on the applications. Itaniums are not slow. The first generation
wasn't really fast at integer operations but flew on fp-Performance (at the
time the first Itanium came out it's fp-Performance outnumbered almost all
CPUs except the POWER4 if I remember correctly). With Itanium2, the integer
performance also made a big step forward.
Performance-wise both Opteron and Itanium are somewhat close, but CPU
performance is just one part of system performance. Opteron for example with
its integrated memory controller is NUMA technology, and in multiprocessor
environments You'll find that the memory performance is very fast if the CPU
needs data from the memory connected to its own memory controller but quite
slow when having to access data residing in memory that's connected to
another CPU.
Besides all technical things You should never forget that You usually don't
buy "an Itanium CPU", "a PA-RISC CPU" or "a UltraSPARC CPU". Usually You buy
a complete package containing hardware and software to solve a certain
problem. Different software runs at different efficiency on different
platforms. Benchmarks like the SPEC can just give You a general overview of
the capabilities, the only thing that counts in professional computing is
the resulting performance on the needed applications.
You don't just buy a CPU but a platform, and with Your platform decision You
also fixate Yourself in terms of service and upgradeability. Sun has it's
market, IBM pSeries (sorry, of course it's p5 now) also has its market, as
have Opteron and the Itanium - despite the bad press. Itanium is a very
powerful platform. We have a bunch of HP Superdomes upgraded from PA-RISC to
Itanium, and we are very satisfied with their crunching capabilities.
Benjamin
Benjamin Gawert
> Sure -- Opteron has very successfully taken the volume market IA64 was
> aiming at. That's why Intel had to do a big U-turn and bring
> Yamhill/EM64T into production, to avoid missing the volume 64 bit
> market completely.
Sure, they realized that putting some 64Bit extensions to a 32bit CPU is
more than enough for the mass market.
>> Not really. The transition from PA-RISC to IA64 was not trouble-free
>> but quite smooth after all. IA64 is the way to go for HP. Even if
>> Itanium doesn't get a high volume chip (which isn't really HPs or
>> intels intention)
>
> Oh it most certainly was, but they now recognise that's not going to
> happen!
When the IA64 project was started years ago it indeed was aimed at the mass
market and as a replacement for x86. However, intel and HP realized very
fast that the Itanium will be too expensive for mass market, and changed
their target at the HPC market and the traditional RISC market (servers and
workstations). This directional change has been made long ago, even long
before the first Itanium(1)s were shipped 1999...
>> it is still way cheaper than the older PA-RISC development. And the
>> new
>
> Itanium is the most expensive CPU ever developed, and by a long way.
Nope. The 1.5GHz Itanium2 for example is still much cheaper per unit than
the (much slower) PA-8800 or even the older PA-8700+, or even an UltraSPARC
III. Of course it's much more expensive than an Opteron CPU, or an Athlon64
or Xeon. But this isn't joe home users market, and the CPU cost are only a
small fraction of the overall system costs. Of course no system vendor gives
up a complete platform that sells at least somewhat ok only because other
CPUs are cheaper.
> Opteron already took almost all IA64's marketplace.
Definitely no. IA64 has a strong stand within the HPC and Highend market
(the market of PA-RISC, SGI MIPS, Sun UltraSPARC III and IBM POWER), and
that's where it has been targeted at since shortly after project start.
Opterons on the other side are clearly aimed at the x86 multiprocessor
market and more competitors of the XEONs instead of Itanium. Perfectly
reflected btw by price and performance of the current mass of opteron
systems.
Benjamin
Andrew Gabriel
"Benjamin Gawert" <bga...@gmx.de> writes:
> Andrew Gabriel wrote:
>
>> Sure -- Opteron has very successfully taken the volume market IA64 was
>> aiming at. That's why Intel had to do a big U-turn and bring
>> Yamhill/EM64T into production, to avoid missing the volume 64 bit
>> market completely.
>
> Sure, they realized that putting some 64Bit extensions to a 32bit CPU is
> more than enough for the mass market.
>
>>> Not really. The transition from PA-RISC to IA64 was not trouble-free
>>> but quite smooth after all. IA64 is the way to go for HP. Even if
>>> Itanium doesn't get a high volume chip (which isn't really HPs or
>>> intels intention)
>>
>> Oh it most certainly was, but they now recognise that's not going to
>> happen!
>
> When the IA64 project was started years ago it indeed was aimed at the mass
> market and as a replacement for x86. However, intel and HP realized very
> fast that the Itanium will be too expensive for mass market, and changed
> their target at the HPC market and the traditional RISC market (servers and
> workstations). This directional change has been made long ago, even long
> before the first Itanium(1)s were shipped 1999...
Sorry, this isn't true. No one invests the amount Intel did in a
chip which isn't for high volume. Intel still believed it would be
a high volume chip until Opteron started shipping in volume over
the last year or so.
>>> it is still way cheaper than the older PA-RISC development. And the
>>> new
>>
>> Itanium is the most expensive CPU ever developed, and by a long way.
>
> Nope. The 1.5GHz Itanium2 for example is still much cheaper per unit than
> the (much slower) PA-8800 or even the older PA-8700+, or even an UltraSPARC
This is only valid if you write off the development costs. You can
only do that for high volume chips (or at least, you can partially
ignore them as they are a small part of the chip price). This isn't
true of Itanium, for which the development costs are the highest of
any chip ever designed. At current volumes, Intel will never recoup
even a small part of their development cost. That will make Itanium
the most expensive mistake in computer history. The production cost
of a low volume chip is not the significant part, but that's all
Intel can currently charge for it or it would have no sales of it.
> III. Of course it's much more expensive than an Opteron CPU, or an Athlon64
> or Xeon. But this isn't joe home users market, and the CPU cost are only a
> small fraction of the overall system costs. Of course no system vendor gives
> up a complete platform that sells at least somewhat ok only because other
> CPUs are cheaper.
>
>> Opteron already took almost all IA64's marketplace.
>
> Definitely no. IA64 has a strong stand within the HPC and Highend market
> (the market of PA-RISC, SGI MIPS, Sun UltraSPARC III and IBM POWER), and
> that's where it has been targeted at since shortly after project start.
That simply isn't true. Intel would not have waited so long to enter
the x86-64 market if it was, because by doing so, it's now playing
catch-up just to become the also-ran, not the leader.
> Opterons on the other side are clearly aimed at the x86 multiprocessor
> market and more competitors of the XEONs instead of Itanium. Perfectly
> reflected btw by price and performance of the current mass of opteron
> systems.
--
Thomas Dehn
Well, for the next few years we still have
- the x86 family (Intel, AMD, and VIA)
- PowerPC (IBM)
- Sparc (Sun, TI, Fujitsu)
Plus a few smaller ones in niche markets such as supercomputers.
We also have billions of processor chips in other devices, such
as mobile telephones, smartcards, and even in cars. As these have
other demands than computers, it is quite possible
that different hardware architectures prosper there.
Thomas
Benjamin Gawert
>> When the IA64 project was started years ago it indeed was aimed at
>> the mass market and as a replacement for x86. However, intel and HP
>> realized very fast that the Itanium will be too expensive for mass
>> market, and changed their target at the HPC market and the
>> traditional RISC market (servers and workstations). This directional
>> change has been made long ago, even long before the first
>> Itanium(1)s were shipped 1999...
>
> Sorry, this isn't true. No one invests the amount Intel did in a
> chip which isn't for high volume.
Oh, it definitely is. At the beginning IA64 was intended to replace x86 but
shortly after the project started intel realized that this wasn't possible.
You clearly forget that Itanium is not only intel's baby. It's a
co-development by intel and HP, and for HP it never was a volume market chip
but a replacement for their PA-RISC processors which also were no high
volume processors. With HP sharing the investment intel realized very fast
that it won't be selling in high numbers but on the other side is a perfect
way to get a foot into the High Performance Computing area.
> Intel still believed it would be
> a high volume chip until Opteron started shipping in volume over
> the last year or so.
Nope. We got our first Itanium systems 1999, and I got my personal machine
mid-2001. Even at that time no-one at intel was talking about Itanium being
a mass-market CPU but a high end solution...
The fact that it also runs Windowsxp (and this in 64bit since 2001!) doesn't
mean it is competing with PCs...
>> Nope. The 1.5GHz Itanium2 for example is still much cheaper per unit
>> than the (much slower) PA-8800 or even the older PA-8700+, or even
>> an UltraSPARC
>
> This is only valid if you write off the development costs.
No. For HP (which had to carry a big bunch of the development costs) the
Itanium is around 40% cheaper than a PA-8800, and this includes the
development costs...
> You can
> only do that for high volume chips (or at least, you can partially
> ignore them as they are a small part of the chip price).
Wrong. Even in low-volume markets the CPU price also carries the development
costs. That's one reason CPUs like MIPS R16000A, Sun UltraSPARC III or
POWER5 are that expensive.
> This isn't
> true of Itanium, for which the development costs are the highest of
> any chip ever designed. At current volumes, Intel will never recoup
> even a small part of their development cost. That will make Itanium
> the most expensive mistake in computer history.
Nope. As I said often enough Itanium isn't a mass-market product, and like
other CPUs that are not mass-market products won't follow the rules You know
from Your average Athlon PC.
I don't see why Itanium should be the most expensive mistake in computer
history. Such statements are pure BS and mostly come from people that never
even touched an Itanium machine (forget about having worked with it). You
won't hear such statements from people with at least a fraction of
competence in the high end area. IA64 had a difficult start but it's still a
great platform for certain applications (as are all the other
non-mass-market platforms like Sun and such). For our company's demands
Itanium on HP-UX gives us the best bang for the buck for our applications.
But then, we buy complete solutions based on a certain platform, and in the
overall machine price the CPU costs are just a small fraction that simply
doesn't matter...
>> Definitely no. IA64 has a strong stand within the HPC and Highend
>> market (the market of PA-RISC, SGI MIPS, Sun UltraSPARC III and IBM
>> POWER), and that's where it has been targeted at since shortly after
>> project start.
>
> That simply isn't true. Intel would not have waited so long to enter
> the x86-64 market if it was, because by doing so, it's now playing
> catch-up just to become the also-ran, not the leader.
intel waited that long simply for two reasons: the lack of 64Bit operating
systems for x86-64, and the fact that most applications usually served by
x86 machines don't really need 64Bit at all. Now, with Linux doing x86-64
quite well and Windowsxp and Windows 2003 server for x86-64 being close to
appear on the scene it finally made sense for intel to participate with
64Bit extensions for their 32Bit CPUs. Especially with more and more PC
kiddies rushing out for a 64Bit Athlon because 64Bit make their games faster
;-)
Benjamin
GreyCloud
Benjamin Gawert wrote:
Thnx for the heads up. A good report.
peder
Actually John, Erik's guess is far closer to reality than your slanted
assumption. There were a number of factors that pushed out the Sol
x86 delivery to be later than Linux. That said, it was always
communicated from day one that JDS was a solution that ran on
Solaris/SPARC, Sol x86 and Linux. I hate to let you down John, but my
coming from Cobalt is irrelevant to the desktop strategy.
>
> If you Google Peder Ulander you'll find a better guess why MadHatter for
> Solaris on the SunRay and Blade 2000 was an after-thought for the
> Cobalt Networks alumni.
JDS was/is actually delivered on Solaris as an integrated set of
applications (GNOME, Mozilla, Evo, Star, etc.) well over a year before
the first packaged JDS for Linux shipped in December. Sun has had
great success with the Sun Rays and Sun Blades running the JDS stack.
The Linux play was an extension to the Sun Desktop strategy, enabling
us to go after CIOs that were already preferring Linux/OSS and
interested in desktop solutions.
-peder
>
> John
> groe...@acm.org
Timothy J. Bogart
Did I get it wrong, or was the real point that an update to the PerlDBI
was needed to get around what was alleged to be a fundamental lack in
Postgres - the lack of cursor support. Including, in the dialog, a
message from Postgres folks indicating it wasn't going to be addressed
anytime soon.
Paul Eggert
>>>http://blogs.sun.com/roller/comments/alanbur/Weblog/elbonian_database_technology
> Did I get it wrong, or was the real point that an update to the
> PerlDBI was needed to get around what was alleged to be a fundamental
> lack in Postgres - the lack of cursor support.
It's kind of hard to tell from the blog. Postgres has cursors, but
there are some restrictions on their use. It didn't appear to me
that his needs would run afoul of those restrictions, but his needs
were stated pretty vaguely and scatteredly so it's hard to tell.
It was clear that the Perl DBI interface to Postgres did some
suboptimal things in response to some of those restrictions, but
that's a different matter.
The quote that caught my attention was:
The reason I originally looked as Postgres was precisely because of
it's good support for SQL99, it's just the Perl DBI interface which
isn't up to the job.
-- Posted by Alan Burlison on July 24, 2004 at 12:32 AM PDT
Rich Teer
> The snapshot of the author in Rich's forthcoming book should be
> really interesting ... :-)
My recently updated website, www.rite-group.com/rich, contains
a picture of my ugly mug (and my dog, Judge). The site also
contains all the info about said forthcoming book, Solaris
Systems Programming, so I encourage everyone to have look and
buy a copy or two!
--
Rich Teer, SCNA, SCSA, author of "Solaris Systems Programming",
publishing in August 2004.
President,
Rite Online Inc.
Voice: +1 (250) 979-1638
URL: http://www.rite-group.com/rich
Erik Magnuson
> > Hazardous guess.
>
> Actually John, Erik's guess is far closer to reality than your slanted
> assumption. There were a number of factors that pushed out the Sol
> x86 delivery to be later than Linux. That said, it was always
> communicated from day one that JDS was a solution that ran on
> Solaris/SPARC, Sol x86 and Linux. I hate to let you down John, but my
> coming from Cobalt is irrelevant to the desktop strategy.
Thanks for affirming my guess. The guess was based on experience with
Linux and Solaris (both Sparc and x86), with Linux having a
substantial lead over Solaris in the x86 world.
My hope is that work on the AMD64 port of Solaris will increase
availability of drivers for both AMD64 and Sparc. This may be made
easier by most add-on devices being either 1394 or USB based rather
than PCI based.
> >
> > If you Google Peder Ulander you'll find a better guess why MadHatter for
> > Solaris on the SunRay and Blade 2000 was an after-thought for the
> > Cobalt Networks alumni.
>
> JDS was/is actually delivered on Solaris as an integrated set of
> applications (GNOME, Mozilla, Evo, Star, etc.) well over a year before
> the first packaged JDS for Linux shipped in December. Sun has had
> great success with the Sun Rays and Sun Blades running the JDS stack.
> The Linux play was an extension to the Sun Desktop strategy, enabling
> us to go after CIOs that were already preferring Linux/OSS and
> interested in desktop solutions.
>
Sun was touting Solaris x86 in 2001 as being suitable for for many
home and small office tasks, using StarOffice and Netscape. Don't
think that passed the giggle test, though. Going back 10 years, HP was
putting quite a bit of effort into making HP-UX suitable for general
desktops, having Lotus 123 ported in late 1991 and AmiPro ported a
couple of years later (this was obviously not for generic PC's).
A lot of companies have been adverse to using non-x86 hardware, even
when a higher performance hardware would pay for itself in a few
months (e.g. a machine to support a $25,000 FEA package). The usual
line is that once the machine is too slow to use as a workstation, it
can then be assigned to a secretary (I have yet to see that happen).
With that kind of mindset, JDS would be an easier sell to CIO's if it
could run on generic PC hardware.
One further Linux vs Solaris comment. Sun has been under a lot of
pressure to abandon Solari in favor of linux (particularly from the
"investment community). What most of these folks don't understand is
that most of what goes into a Linux distro can be ported to Solaris
without too much difficulty.
Alan Coopersmith
|JDS was/is actually delivered on Solaris as an integrated set of
|applications (GNOME, Mozilla, Evo, Star, etc.) well over a year before
|the first packaged JDS for Linux shipped in December.
Except for Evolution, which Sun has never released for Solaris, and for
the integration between the other ones, which isn't really much to speak
of. (You can't even save your Mozilla windows as part of your GNOME
session, since they don't speak the same session saving protocol.)
--
________________________________________________________________________
Alan Coopersmith * al...@alum.calberkeley.org * Alan.Coo...@Sun.COM
http://www.csua.berkeley.edu/~alanc/ * http://blogs.sun.com/alanc/
Working for, but definitely not speaking for, Sun Microsystems, Inc.
GreyCloud
Rich Teer wrote:
> On Mon, 26 Jul 2004, Chris Thompson wrote:
>
>
>>The snapshot of the author in Rich's forthcoming book should be
>>really interesting ... :-)
>
>
> My recently updated website, www.rite-group.com/rich, contains
> a picture of my ugly mug (and my dog, Judge).
Hey, don't be hard on yourself. Nice dog too and nice displays.
> The site also
> contains all the info about said forthcoming book, Solaris
> Systems Programming, so I encourage everyone to have look and
> buy a copy or two!
>
--
Rich Teer
> Hey, don't be hard on yourself. Nice dog too and nice displays.
Thanks; I don't know how I'd cope without Judge and my twin
21" Sun monitors!
As for the former, how about "My dashing good looks can be
seen from my new website"? :-)
Arthur Corliss
On 2004-07-31, Thomas Dehn <thomas...@arcor.de> wrote:
Bdd
>
> Well, for the next few years we still have
> - the x86 family (Intel, AMD, and VIA)
> - PowerPC (IBM)
> - Sparc (Sun, TI, Fujitsu)
> Plus a few smaller ones in niche markets such as supercomputers.
> We also have billions of processor chips in other devices, such
> as mobile telephones, smartcards, and even in cars. As these have
> other demands than computers, it is quite possible
> that different hardware architectures prosper there.
Don't forget the MIPS family, the foundation of the SGI/IRIX platform.
There are certainly less choices than there used to be, but there's still
a lot of competitive innovation going on between the remaining players.
--
--Arthur Corliss
Bolverk's Lair -- http://arthur.corlissfamily.org/
Digital Mages -- http://www.digitalmages.com/
"Live Free or Die, the Only Way to Live" -- NH State Motto
Benjamin Gawert
> Don't forget the MIPS family, the foundation of the SGI/IRIX platform.
Well, SGI is working on getting rid of them and concentrating on Itanium and
Linux, so I wouldn't take the MIPS/IRIX systems to have somewhat like a
future. No wonder when looking on the low performance of their current MIPS
line...
Benjamin
JEDIDIAH
On 2004-07-31, Benjamin Gawert <bga...@gmx.de> wrote:
> Andrew Gabriel wrote:
>
>> Sure -- Opteron has very successfully taken the volume market IA64 was
>> aiming at. That's why Intel had to do a big U-turn and bring
>> Yamhill/EM64T into production, to avoid missing the volume 64 bit
>> market completely.
>
> Sure, they realized that putting some 64Bit extensions to a 32bit CPU is
> more than enough for the mass market.
Well, if you design the 32bit processor right you will only need "extensions"
in order for it to support larger operands. The 64bit processor shouldn't
be doing things fundementally different from the 32bit ones.
[deletia]
--
It is not true that Microsoft doesn't innovate.
They brought us the email virus.
In my Atari days, such a notion would have |||
been considered a complete absurdity. / | \
Leon Workman
> > SUN is going in the complete opposite direction. It is blindly
> > clinging on to Solaris instead of switching to Linux OR having a Linux
> > affinity Solaris such as something that could be branded "SolarisLX"
> > that would work seemlessly with Linux. Make Linux or Linux affinity
> > Solaris availible for everything from a SUNFIRE to the Ultrasparc.
>
> Allow me to be blunt: you're trippin'.
>
> Why would Sun dump a reliable, robust OS whose functionality Linux has yet
> to achieve? That would not be prudent, to say the least. Some of Sun's
> biggest customers are largest banks in the world, and when you mention Linux
> to them, their hair stands up on their heads (first hand experience).
> That's just one example. There are many.
> Linux isn't even close in robustness and functionality of Solaris, and it'll
> be "a while" before it gets there. But who knows how far ahead will Solaris
> be then?
I've been seeing an alarming trend in the marketplace with companies
that embrace Linux (with one of them, at least). HP was one of the
first vendors to jump on the Linux bandwagon (and Microsoft
interoperability bandwagon nearly simultaneously), and their unix
isn't what it once was. They had to cut costs somewhere due to the
slump in the economy, and I believe that unix and unix middleware R&D
took some real hits. The company was more focused on Linux (and
printing and digital photography and Microsoft garbage and scooping up
other flailing companies) than its own OS, and as a result, they've
lagged a bit behind their competition.
To embrace Linux at the expense of wasting all of your R&D $'s that
you've already put into a more reliable OS is just plain retarded.
Sure, Linux is cheap. Sure, Linux seems to do a whole hell of a lot
better than Windows in the reliability area, but it's all open source.
All of these companies are eating up what amounts to freeware,
putting their sticker on it, and reselling it without even knowing
what a lot of the underlying code is. It's going to bite them hard
somewhere in the future. Not everyone that contributes freeware has
altruistic intentions. Some disgruntled troll somewhere has probably
already put the next Y2K into some obscure Linux code.
Nothing's free.
Stepping off of my soap box...
Leon
GreyCloud
Rich Teer wrote:
> On Sun, 1 Aug 2004, GreyCloud wrote:
>
>
>>Hey, don't be hard on yourself. Nice dog too and nice displays.
>
>
> Thanks; I don't know how I'd cope without Judge and my twin
> 21" Sun monitors!
>
> As for the former, how about "My dashing good looks can be
> seen from my new website"? :-)
>
:-))
I could certainly not brag about my looks.
I could break a mirror with my looks.
That's why I don't look in them.
Erik Magnuson
>
> My recently updated website, www.rite-group.com/rich, contains
> a picture of my ugly mug (and my dog, Judge). The site also
> contains all the info about said forthcoming book, Solaris
> Systems Programming, so I encourage everyone to have look and
> buy a copy or two!
Surprisingly enough, my monitor survived the strain of rendering
the image of your mug - might have to do something with it being
seismically reinforced to withstand the So Cal earthquakes...
Nice use of whitespace in your website. One thing I did miss was
a link to the site's homepage from the underlying links - my
suggestion would be embedding it in the site logo (click on the
logo to get back to home).
Looking forward to your book as well and will be bugging San
Diego Tech Books to stock it.
-Erik
Rich Teer
> Surprisingly enough, my monitor survived the strain of rendering
> the image of your mug - might have to do something with it being
> seismically reinforced to withstand the So Cal earthquakes...
Yeah, that would do it! :-)
> Nice use of whitespace in your website. One thing I did miss was
> a link to the site's homepage from the underlying links - my
> suggestion would be embedding it in the site logo (click on the
> logo to get back to home).
Thanks for the feedback. The old site used to have that feature,
and I agree that the new one should too. I've got family visiting
from the UK for the next 4 weeks or so, so any changes probably
won't appear until they've gone home.
> Looking forward to your book as well and will be bugging San
> Diego Tech Books to stock it.
Much oblidged! I hope the book's web site contains all the info
you need (see the comment above for an indication of when the
index will appear).
John D Groenveld
peder <peder....@gmail.com> wrote:
>assumption. There were a number of factors that pushed out the Sol
>x86 delivery to be later than Linux. That said, it was always
Besides the killing of StarOffice for Solaris x86 by Desktop Software
in 2003, what other factors pushed out Solaris x86 delivery later
than Linux?
>JDS was/is actually delivered on Solaris as an integrated set of
>applications (GNOME, Mozilla, Evo, Star, etc.) well over a year before
>the first packaged JDS for Linux shipped in December. Sun has had
What are the differences between the Enterprise Linux Client stack
for Solaris delivered over a year before the first packaged JDS for
Linux shipped in December and Desktop Software's Project Wells?
John
groe...@acm.org
Arthur Corliss
>
> Well, SGI is working on getting rid of them and concentrating on Itanium and
> Linux, so I wouldn't take the MIPS/IRIX systems to have somewhat like a
> future. No wonder when looking on the low performance of their current MIPS
> line...
;-) Don't you have some anti-Linux/x86 zealots to poke in comp.sys.sgi.*?
I disagree with how much engineering SGI has put into MIPS, to be sure, but
it's still a much better architecture than x86/VILW crap. MIPS has been
pretty consistent over the years in getting more ops executed per clock
cycle than just about everyone, including RISC vendors. Great family, it's
just unfortunate that SGI has chosen to not push it like they should have.
And don't diss IRIX too much -- like AIX, IRIX is another OS that offers a
lot of features that I wish Linux would get. Linux is a good OS, but it has
a long ways to go before it can rival either of them in functionality and
robustness.
peder
> In article <1b1456cc.04073...@posting.google.com>,
> peder <peder....@gmail.com> wrote:
> >assumption. There were a number of factors that pushed out the Sol
> >x86 delivery to be later than Linux. That said, it was always
>
> Besides the killing of StarOffice for Solaris x86 by Desktop Software
> in 2003, what other factors pushed out Solaris x86 delivery later
> than Linux?
Sun never "killed" Star for Sol x86. At the point in time that
StarOffice 7 went into development and testing, Sol x86 was in it's
"indefinitely delayed" period. When Sol x86 was put back on the
priority list, the engineering team jumped on creating a supported
Star 7 package. It takes more than a day or a couple of weeks to
build and test software.
The factors that actually had a huge impact on JDS Sol x86 include
HCL, drivers for cameras, printers, graphics cards, etc. The most
important thing that had impact here was that there was no support for
the latest versions of Mozilla, Evolution, GNOME, etc. in Sol x86 -
meaning that Sun needed to develop and manage the ports. All of this
stuff being readily available in Linux, gave the Linux kit a faster
TTM. Rather than holding the release up of JDS v1, we launched it to
make sure that we had first mover advantage in the alternative
enterprise client space.
> >JDS was/is actually delivered on Solaris as an integrated set of
> >applications (GNOME, Mozilla, Evo, Star, etc.) well over a year before
> >the first packaged JDS for Linux shipped in December. Sun has had
>
> What are the differences between the Enterprise Linux Client stack
> for Solaris delivered over a year before the first packaged JDS for
> Linux shipped in December and Desktop Software's Project Wells?
It was an enterprise client stack for Solaris (not Linux) that was
first introduced a while back. It ran GNOME 2.0, Mozilla, Star and
there is a beta version of Evo that was available. Wells is the
updated integration of the tools.
-peder
Joerg Schilling
peder <ula...@gmail.com> wrote:
>groe...@cse.psu.edu (John D Groenveld) wrote in message news:<cep2o1$n18$1...@neuromancer.cse.psu.edu>...
>> In article <1b1456cc.04073...@posting.google.com>,
>> peder <peder....@gmail.com> wrote:
>> >assumption. There were a number of factors that pushed out the Sol
>> >x86 delivery to be later than Linux. That said, it was always
>>
>> Besides the killing of StarOffice for Solaris x86 by Desktop Software
>> in 2003, what other factors pushed out Solaris x86 delivery later
>> than Linux?
>
>Sun never "killed" Star for Sol x86. At the point in time that
Correct - they cannot ;-)
Star is available since 1985 (at this time only for SunOS-2.x) I did never
announce a kill for any of the of the supported platforms...
Source:
ftp://ftp.berlios.de/pub/star/
ftp://ftp.berlios.de/pub/star/alpha/
Solaris binary install packages:
http://www.blastwave.org/packages.php
--
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
John D Groenveld
It looked that way from customers' perspective: no longer supporting
the Solaris x86 version of the Star Office 6.0 product by distributing
patches in 2002 and then not including a Solaris x86 build in the
StarOffice 6.1 (7) Early Access program in 2003, well after Solaris x86
was resurrected.
Was that "false" impression Desktop Software Marketing's or Curtis
Sasaki's [1] bug?
>The factors that actually had a huge impact on JDS Sol x86 include
>HCL, drivers for cameras, printers, graphics cards, etc. The most
>important thing that had impact here was that there was no support for
>the latest versions of Mozilla, Evolution, GNOME, etc. in Sol x86 -
Did Geoff Baysinger mis-speak when at SunNetwork 2003 in San Francisco
he explained that the lack of support for StarOffice for Solaris x86
was the "principle problem" in getting JDS for Solaris x86?
>It was an enterprise client stack for Solaris (not Linux) that was
>first introduced a while back. It ran GNOME 2.0, Mozilla, Star and
>there is a beta version of Evo that was available. Wells is the
>updated integration of the tools.
How did John Loiacono get the "false" impression [2] that the ELC product
that Desktop Solutions was hawking was actually Enterprise Linux Client?
If the client stack has long been available for Solaris (SPARC)
why did Jonathan Schwartz fail to mention that it ran at the time
on Sun's SunRay systems when he demo'd the client stack on that
Linux white box at the Project MadHatter unveiling at SunNetwork 2002?
Why does Dan Baigent's December 11th SJDS 2003 Technical Overview
presentation list Sun Ray Server for Linux as higher confidence for
CY2004 than Project Wells: "Solaris SPARC Only!" ?
Why is JDS for Solaris (SPARC) and the Sun Ray still not available
for purchase from SunStore [3]?
1. <URL:http://www.save-solaris.org/sasaki.html>
2. <URL:http://www.theregister.co.uk/2003/03/29/suns_linux_distro_is_dead/>
| Interestingly, John Loiacono, vice president of operating platforms at
| Sun said that the company now defines the ELC (Enterprise Linux
| Client) as "software" rather than "hardware". A hint that Sun really,
| really doesn't want to be in the box-shifting business when the Client
| launches in the middle of this year.
3. <URL:http://store.sun.com/CMTemplate/Catalog/PartDescPopup.jsp?pm_id=3b1d04:fe2ef9c06f:-5d26&FamOID=4AD8A261D21DB211AF6DD75BED4369F8>
| Operating System Java Desktop System includes the operating
| environment, which is Linux.
John
groe...@acm.org
Benjamin Gawert
> I disagree with how much engineering SGI has put into MIPS, to be
> sure, but it's still a much better architecture than x86/VILW crap.
In what regards is MIPS better than x86 or IA64 (I assume You are targeting
Itanium with "VLIW")?
> MIPS has been pretty consistent over the years in getting more ops
> executed per clock cycle than just about everyone, including RISC
> vendors.
Well, MIPS _is_ RISC. And besides that, Your statement is wrong. POWER,
PA-RISC and Itanium do more OPS/clock cycle than even the R16k.
And not to forget that this is just a theoretical view. The only thing that
counts is real world performance. It doesn't count if MIPS does more
instructions per clock cycle if the absolute system performance is behind of
all the competitors...
> Great family, it's just unfortunate that SGI has chosen to
> not push it like they should have.
SGI made a wise decision to look for something different. Something that's
more up to date. Something SGI can make some money. Something software
vendors are willing to write software for (which isn't the case for IRIX).
Looks they are very successful with their Itanium-based Altix...
Benjamin
Timothy J. Bogart
I guess I said that badly because I wasn't paying enough attention.
You are right, there is cursor support of sorts, but the author contacts
the postgress guys with this linked correspondence:
"Leon Brocard wrote:
>Hi, I need to iterate over millions of rows in a few tables. Does the
>text at:
>
> http://search.cpan.org/dist/DBD-Pg/Pg.pm#Cursors
>
>still hold? If I understand this, this basically means that I can't
>use pg for my needs. Is there any chance of better cursors soon? Is
>there a technique to work around this?
>
>Cheers, Leon
>
>
You can iterate over large sets in batches using SELECT queries with
LIMIT and OFFSET clauses inside a loop. I don't know if this will help
you, but I have used it in the past.
Work is being committed to the postgress core at this moment to provide
for nested transactions, so cursors might be more useful in future, but
for now you need to take care with large data sets.
cheers
andrew"
Though I think the fundamental issue remains the same - the two OSS
databases tried (MySql and Postgress) do not handle cursors according to
his needs. It would seem this sort of thing is customary on commercial
databases.
I wish I were a bit more up on the actual coding involved ...
peder
> Sasaki's [1] bug?
Before my time, so I can't tell you if it was Curtis's or Marketing.
Besides, it is irrelevant at this point as the product is there and
shipping in 7 today. On top of that, a Sol x86 version has been part
of the OpenOffice community through the 1.1, so baseline development
and testing was done by the Sun team (with help from the community).
>
> Did Geoff Baysinger mis-speak when at SunNetwork 2003 in San Francisco
> he explained that the lack of support for StarOffice for Solaris x86
> was the "principle problem" in getting JDS for Solaris x86?
Yes he mis-spoke. If StarOffice x86 was the key factor causing delay,
don't you think that you would see JDS on Sol x86 today? There is
still a lot of work happening on the OSS applications and HCL. At
this point, I would suggest you focus your efforts on getting a Sol
x86 port of Evolution, Mozilla, etc. as well as get the necessary
drivers in place as that would go a long way in shortening the
development cycle
>
> How did John Loiacono get the "false" impression [2] that the ELC product
> that Desktop Solutions was hawking was actually Enterprise Linux Client?
I'm not sure I'm understanding what you are saying. We launched a
desktop client that was based on Linux, specifically targeted at the
enterprises that we looking/commited to Linux. We had, even at this
time, GNOME, Mozilla, Star running on Solaris. What is the point that
you are trying to make? We had a Linux version and a Solaris version.
The article you refer to specifically is discussing Sun's decision to
abandon Sun Linux.
>
> If the client stack has long been available for Solaris (SPARC)
> why did Jonathan Schwartz fail to mention that it ran at the time
> on Sun's SunRay systems when he demo'd the client stack on that
> Linux white box at the Project MadHatter unveiling at SunNetwork 2002?
You are talking about a launch message. At the MadHatter launch,
JIS's speech was specifically in place to launch a turnkey, blackbox
enterprise client that would extend Sun into the emerging
"Linux-preferred" market. At the same keynote, JIS did demonstrate
the Sun Ray in that demo (with the enterprise client) as the Java Card
was a core feature for security that was also launched at SNC 2002.
>
> Why does Dan Baigent's December 11th SJDS 2003 Technical Overview
> presentation list Sun Ray Server for Linux as higher confidence for
> CY2004 than Project Wells: "Solaris SPARC Only!" ?
Similar to my earlier statement, your help in getting the latest
versions of GNOME, Mozilla, Evolution, etc to Solaris (SPARC and Sol
x86) would go a long way in getting these products to market a lot
faster.
>
> Why is JDS for Solaris (SPARC) and the Sun Ray still not available
> for purchase from SunStore [3]?
The full productization is still under development because of lack of
OSS application support (GNOME 2.x, Evolution 2.x, rDesktop, GIMP,
etc.) The other components they come for free for Solaris. They also
ship with the workstations and the Sun Ray's, so there is no need to
purchase separately on the store.
http://wwws.sun.com/software/star/gnome/get/index.html
http://wwws.sun.com/software/solaris/browser/
Ximian is available through Novell:
http://rpmfind.net/linux/RPM/ximian/ximian-evolution/solaris-8-sun4/
And StarOffice for Solaris x86 is available through the store and from
OpenOffice.org
-peder