After being promoted to the No. 2 position at Sun Microsystems, Jonathan Schwartz has begun spreading his unusual pricing plans from the software group to the rest of the company.
Schwartz Spreads Pricing Ideas Beyond Sun Software
About The Author
Eugenia Loli
Ex-programmer, ex-editor in chief at OSNews.com, now a visual artist/filmmaker.
Follow me on Twitter @EugeniaLoli
“We’re beginning to put some meat on the bones of the strategy around using both technology and economics to drive disruptive innovation in the enterprise,” Schwartz said in an interview.
I don’t know about you, but I was impressed with the buzzwords, “disruptive innovation”.
Sun’s name and disruptive technology are an oxymoron.
They were the last vendor to embrace Linux and only half-heartedly. They do not realize that the world of expensive proprietary boxes tied to expensive proprietary hardware is over and gone and that the future belongs to small, increasingly-more powerful devices running Linux.
All of Sun’s engineering will not save them. By the time they are done plannnig, doing prototypes of the next big thing, the world will have left them behind. The only thing that may save them is if they begin to offer SunRay’s technology at a very, very cheap price.
And arguably, this is a technology that is too server centric, so they will only score some goals there.
Good luck, dinosaurs. You have already earned your place in the museum of computing. No small feat, yet a sad ending.
I don’t care if he’s a founder, he has presided over a 90% drop in the stock price, and the company continues to be totally rudderless and almost incredulous with respect to market conditions – witness the talk last week of a renewed push for Solaris x86. Solaris x86? Talk about a zombie product. The fact that the company would even let whispers surrounding this dud product swirl about clearly indicates the “Helter Skelter” mindset McNealy is in.
Shareholders should put his head on a stick and figure out how this company is going to survive.
> Sun’s name and disruptive technology are an oxymoron. They were the last vendor to embrace Linux and only half-heartedly.
Since when Linux started being an indicator of inovation, did I miss a memo or something? There is little if any innovation in Linux, it is just a cheap carbon copy of other technologies developed by other companies that really innovate (Sun springs to mind). Linux is Unix clone by definition with other technologies slapped on top. If Sun didn’t embrace Linux wholeheartedly, that doesn’t mean that Sun stopped innovating Solaris, which is a better, more full-featured, and more stable OS than Linux as it is. After Solaris 10 comes out Linux will be in the dust as far innovation is concerned.
They were the last vendor to embrace Linux and only half-heartedly. They do not realize that the world of expensive proprietary boxes tied to expensive proprietary hardware is over and gone and that the future belongs to small, increasingly-more powerful devices running Linux.
Words of a true Linux troll.
And arguably, this is a technology that is too server centric, so they will only score some goals there.
In case you haven’t noticed, Sun has always been about servers and high end workstation. They’re not in the “hey we release a distro with 8 office suites and 20 webbrowsers and 60 e-mail clients”. They’re in the “doing it right” business.
This ones a bit more broad. Talks about new subscription pricing, Sun’s financial situation, and Schwartz’s future plans.
http://news.ft.com/servlet/ContentServer?pagename=FT.com/StoryFT/Fu…
Hey anonymous,
Why do you need to hide behind an anonymous proxy? Is it so that you can post twice under the pretense of being two separate persons?
You should follows the development of the Linux kernel a bit more closely to understand innovation.
Pray tell me, which OS has better and more extensive driver support Solaris for X86 or Linux?
How many platforms does Solaris run on? Answer 2 and in one of them very poorly. Go and check how many platforms Linux supports.
Finally, you obviously do not understand how Linux is changing the economics of software development and the effect that it is having in competitiveness. Don’t worry, you’ll realize soon enough…
You may not like what Eu said, but it is in fact true to some degree.
To give an example, I began working at a company a couple of years ago. When I started, they had several Sun servers, a bunch of Sun workstations, and a couple of Irix machines. Two years later, they have 0 Sun machines and 0 Irix machines (in all fairness, they still use the monitors though). In their place, they have several blade servers running Linux and a bunch of dual CPU workstations running Linux (again, to be fair, there are also a couple of OpenBSD machines).
This is not an isolated situation either. If it were, Sun would be doing a lot better than they currently are, wouldn’t they?
Well, it can be that the hardware support of Linux is better than that of Solaris. But then, does that make Solaris a bad OS?
I mean, Mac OS X runs on only one platform. Darwin, the core, also runs on Intel but supports less hardware on it than Linux. Now does that mean that Linux is great and OSX is crap?
> Two years later, they have 0 Sun machines and 0 Irix machines
It makes no sense to extrapolate your small time encouters to overall state of the company. Just to give you an example, the company I’m working with is bringing in 150+ Sun servers on top of already existing ones for its internet architecture. The number of Linux machines brought in is zero, zilch, nada. Why? Because Linux is not stable enough, maintainable enough, or mature enough for this scale of deployment. Sorry to blow your bubble, but Sun machines are still being bought and will be bought in rather large numbers because they can really do the job and be actually cheaper in the long term than Lintel.
Err little or no innovation in Linux? A cheap carbon copy?
I suppose you think Solaris is the pinnacle of operating systems do you?
Linux is far lighter weight in basic operations than Solaris is. Compare system calls, context switching, process creation, network overhead, etc. Linux will hands down beat Solaris, yep try it on SPARC hardware too, Linux still wins.
Oh but you’ll say, “Solaris may trade off a bit at the lowend, but just put it in a big system and it will blow the doors of Linux”? Actually no, SGI has got the old Linux 2.4 kernel running on a 512 CPU shared memory system. They are hoping to have 2.6 handle that pretty much out of the box, and even go up to 1024 CPUs.
I assume you’re going to tell me that NASA doesn’t actually do anything useful with this system, so they don’t care that it isn’t robust. In fact, they probably just use it to heat a building, right?
Time to pack your bags, old timer.
It makes no sense to extrapolate your small time encouters to overall state of the company. Just to give you an example, the company I’m working with is bringing in 150+ Sun servers on top of already existing ones for its internet architecture.
Umm, are you stupid or something? Why do you think your “small time encouters [sic]” are any more worth extrapolating?
You know what to do? Look at the company’s public financial disclosures. I think you’ll find that if Sun’s trend over the last few years continues, it does not face a rosy future.
> Actually no, SGI has got the old Linux 2.4 kernel running on a 512 CPU shared memory system. They are hoping to have 2.6 handle that pretty much out of the box, and even go up to 1024 CPUs.
This is a totally pointless comparison. SGI Altix systems are NUMA, whereas Sun systems are SMP, which makes it an apples to oranges comparison. NUMA and SMP are targeted at totally different application domains (high memory latency / low memory latency). NUMA will always be useless in the area of enterprise/business computing, this is exactly why SGI does not even attempt to market Altix in the enterprise — it is a very good number cruncher for science and visualization, but it is not very useful in the enterpirse. Plus from engineering standpoint building machines with a high number of processors is a lot easier with NUMA than SMP. On the same token making an OS work under NUMA with a large number of processors is not that much of big deal compared with SMP, it just so happend that SGI chose Linux for that particular purpose to reduce the development costs.
Sun systems are not NUMA, you say? So they are a uniform memory architecture? Meaning all CPUs are the same logical distance from all parts of memory, right?
And just which Sun systems are we talking about here? The Sun Fire (E12K, E15K, E20K, E25K) range?
Well that shows how much you know then. Sun’s systems actually have a quite deep NUMA heirachy. Access to different parts of memory is definitely non uniform.
Let me gather some data from Sun’s whitepaper for you.
Memory bandwidth
Access type GB/s
Same CPU 9.6
Same board 6.7
Different board 2.4
Memory latency
Access type ns
Same CPU 180
Same half of board 193
Other half of board 207
Different board 333 to 453
Now you better run off to your boss and tell him to cancel this shipment of useless NUMAised Sun servers.
Oh, the reason Altix is primarily for number crunching and not for “enterprise/business computing” is because the I2 isn’t very strong on integer performance (although better than the likes of what Sun is offering).
> Well that shows how much you know then. Sun’s systems actually have a quite deep NUMA heirachy.
Dude whatever you’re smoking must be some pretty strong stuff, all Sun high end servers are classified as SMP and by no means as NUMA.
> Oh, the reason Altix is primarily for number crunching and not for “enterprise/business computing” is because the I2 isn’t very strong on integer performance
Wrong, even though Itanium’s performance envelope is heavily biased toward floating point performance, its integer performance is still quite decent. HP is using Itaniums in its Superdome server without any drama related to integer performance.
Dude whatever you’re smoking must be some pretty strong stuff, all Sun high end servers are classified as SMP and by no means as NUMA.
Get your hand off it. NUMA. Non uniform memory architecture. WTF does “no means as NUMA” mean? It either has uniform access to memory or it doesn’t. Is that so hard to comprehend? The Opteron is classed as NUMA, and it has a pretty small difference between local and remote memory.
Oh, and Sun’s latency figures are actually pretty comparable to the Altix for your information.
Wrong, even though Itanium’s performance envelope is heavily biased toward floating point performance, its integer performance is still quite decent.
You’re the one smoking crack. When an Opteron or Xeon beats the top end Itanium 2 by 50% for less than half the price, I don’t class it as still quite decent.
HP is using Itaniums in its Superdome server without any drama related to integer performance.
The other reason it is being used in large systems is because its big caches help the interconnect scale to more CPUs.
Notice something?
Because Linux is not stable enough, maintainable enough, or mature enough for this scale of deployment. Sorry to blow your bubble, but Sun machines are still being bought and will be bought in rather large numbers because they can really do the job and be actually cheaper in the long term than Lintel.
I’m glad you think that, but there are many thousands of people who are finding that perception to be wrong. The proof? Sun is going bust, and has been in the red for many, many quarters. Sun machines are not being bought in large enough numbers.
You don’t have any Linux servers in your organization, so without detailing the stability problems this statement means absolutely nothing.
Lintel is also not the only option. IBM sells PPC Linux servers for larger, more specialist ‘enterprize’ grade situations, as do SGI. Sun and Solaris are not low-volume and cheap enough (may have changed with the Opteron, but the opportunity has been lost), and they offer nothing compelling at the specialist computing end any longer.
The situation does not look good, and what is worrying is the number of buzzwords coming out of Sun. Do people like Jonathan Schwartz know how dire the situation is? Pricing plans will do nothing. They need a response to these two assaults, and drive home the message to people out there.
Lintel is also not the only option. IBM sells PPC Linux servers for larger, more specialist ‘enterprize’ grade situations, as do SGI. Sun and Solaris are not low-volume and cheap enough (may have changed with the Opteron, but the opportunity has been lost), and they offer nothing compelling at the specialist computing end any longer.
Exactly what is it Sun isn’t offering that any one else does?
They have Solaris, they have Sparc, they have Linux, they have clusters, they have thin clients, they have workstations, they have it all…
Open your eyes, Sun has what everyone else has AND more… go figure…
> The Opteron is classed as NUMA, and it has a pretty small difference between local and remote memory.
You’re an illiterate moron, buddy. What the hell did they teach you at school? CPU architecture itself can not be automatically classified to some memory access class (NUMA, SMP, etc.). The same CPU architecture can theoretically be used in any architectural class.
> Oh, and Sun’s latency figures are actually pretty comparable to the Altix for your information.
Memory latency is not a real issue here. Enterprise applications (databases especially) are primarily spaghetti code with a lot of branching and therefore requires a significant degree of cache locality for optimum performance. Accessing remote memory is very expesive in NUMA and CC-NUMA in particular, thus NUMA will be considerably less efficient than SMP working on the same workload executing a tipical enterprise application — the performance of CC-NUMA machines is severely limited if data is incorrectly placed at physical memory locations away from the nodes that use it most, if data needs to migrate from one set of nodes to another, or if an application requires a large amount of data to be replicated on each processor.
SUN lately has two faces. One that says Solaris on x86 is great and go Linux, the second is, we hate Linux and open source and we’re gun-ho Solaris, lets load Java on every desktop.
Confusing messages they’re sending out if you ask me. Its like make up your mind already SUN.
Drop the banter about NUMA and SMP! It’s all about plain and simple value for money.
Back in 1998, loads of dot-coms were blowing their investment cash on snazzy Sun kit that looked good in the office/server room, mostly because clueless CEOs/CTOs wanted “real” UNIX, and partly because reselling Sun kit to customers could make them a bit of extra cash.
Fast forward to today and you’ll have noticed that most of these dot-coms are gone. The IT sector quickly regained the clue picked up on about seven or eight years ago that the UNIX experience could be had for a lot less than a bunch of Solaris/SPARC workstations/servers. Sun milked the dot-com game as much as they could, buying themselves a few years, but now normal service has been resumed.
Deny it if you like, but Linux started to change the nature of the game a while back. Sun has a lot of interesting stuff that they could well do to focus on, but without a coherent Linux strategy they’ll soon be axing that interesting stuff or selling it off in order to flog whatever dead horse McNealy and Schwartz have decided to bet the business on.
> SUN lately has two faces. One that says Solaris on x86 is great and go Linux, the second is, we hate Linux and open source and we’re gun-ho Solaris
I don’t think Sun ever had the “we hate Linux and open source” face. Sun is the second biggest contributor to open source after UCB after all. And I don’t thinks there is anything wrong with Sun being “gun-ho Solaris” as you succinctly put it, Linux is a competitor to Solaris in some areas and it should be expected from Sun to emphasize its own OS over Linux just as IBM still tries to push AIX over Linux and HP pushes HP-UX over Linux when possible.
How short everyones memories seem to be. Sun tried to drop the x86 product and there was such an outcry from their customers and the Sun community that they were forced to continue releasing it.
[i]SUN lately has two faces. One that says Solaris on x86 is great and go Linux, the second is, we hate Linux and open source and we’re gun-ho Solaris, lets load Java on every desktop.
Confusing messages they’re sending out if you ask me. Its like make up your mind already SUN.>/I>
I guess you’re not very good at reading news then. Sun isn’t against open source and has never really been. However they DO say that Solaris is a superior solution (which technically most would agree it is). It’s not a bit confusing, they’ve been clear about this issue all the time. They also say that Linux might be a viable desktop option but will support it for servers even though they advice you to use Solaris. Since customer demand has risen for Solaris on x86 they provide that as well.
What exactly is unclear? Is anything that isn’t “We’ll only do Linux and GPL” unclear???? Sun has been clear about this all along and the only thing unclear is the comment section filled with zealotry and FUD about SUN.
I thought Linux people liked certain complexity but I guess I was wrong to a degree. Sun is just listening to the market (as in those with cash not those who scream the highest on /.).
‘m glad you think that, but there are many thousands of people who are finding that perception to be wrong. The proof? Sun is going bust, and has been in the red for many, many quarters. Sun machines are not being bought in large enough numbers.
Wrong again. Sun machines are being bought at larger numbers than before. Notice IDC reports on how sun’s server volumes have increased 26% year on year. They just aren’t getting enough margins on thier sales. Your proof is inadequately researched and wrong.
Lintel is also not the only option. IBM sells PPC Linux servers for larger, more specialist ‘enterprize’ grade situations, as do SGI. Sun and Solaris are not low-volume and cheap enough (may have changed with the Opteron, but the opportunity has been lost), and they offer nothing compelling at the specialist computing end any longer.
Sun has SPARC/Intel servers from $2000 and up. How much cheaper did you expect?
From your posts here it looks like all you know about stuff is the FUD linux zealots spread, go and do some real research before you post.
remind me of the times before the dot.com bubble burst (like 7~8 years ago). Now they sound empty, and slightly ridiculous. It doesn’t feel like Schwartz is in touch with the market.
Sun was a technology leader back then. Now it has lost its edge, its drive, its leader and its market share. They are downsizing fast to avoid a financial disaster.
Sun still has a lot going for it, but it would take more than empty buzzwords to set it on the right path:
1) Leadership. And I am not talking about either McNealy or Schwartz. Better bring somebody from outside, but with enough experience in this industry, and freedom to rebuild a top team.
2) A clear market target. Trying to get a straight answer about where Sun is going these days from a Sun employee has become a challenging task.
3) Shaking out some of the old brass. These guys have been there for too long. Cut the fat.
You kids always have the answer for running (or ruining) every body else’s business. Why don’t you start your own Linux bizness and see if you even get to the $1Million mark.
Sun’s facing problems just like everybody else – SGI, HP, IBM in their “UNIX” server lines.
Why don’t you compare how IBM’s AIX divisionis doing against Dell’s Lintel business?. If you saw the numbers you’d be giving advice to Sam (Palisimo) to shutdown its AIX/PowerPC division and just focus on Lintel.
Compare HP’s HPUX division against IBM’s Linux/Intel division and you’ll be giving Carly to shut down HP’s PARISC/HPUX/Itanium division.
Now compare Sun’s Solaris x86 on AMD vs SGI’s Linux on Itanium and guess what, SGI looks positively horrid. Perhaps you guys should tell Robert (bishop) to shutter SGI.
A lot of proud companies and engineers have this not-invented here attitude and guess what, Linux is not invented here. They have no control on the kernel or the OS.
Only companies that don’t have an OS dept like Dell or IBM’s PC division will opt for something like Linux.
“Only companies that don’t have an OS dept like Dell or IBM’s PC division will opt for something like Linux.”
you have proved your brightness with a comment that ibm doesnt have a os department
IBM PC div uses MS Windows and Linux to flog their x86 boxes. AIX strictly belongs to the RS6K div and s390 has their own OS390/zOS.
Kiddo, get a clue – go browse IBM.com before you come here to debate me
(hint: http://www.pc.ibm.com/us/eserver/xseries/ and look at the two operating systems offered: Linux and Windows).
IBM works on linux and aix alteast which are operating systems. they have a OS department. they have a ltc center which works only on the linux kernel. aix is being developed is what is called as the computing lab. i have been to both of them
bye
“[…] IBM […] HP […]”
Invalid compare: those 2 companies don’t have financial problems. SGI and SUN have, while SGI has already ”pretty much” abandoned IRIX and MIPS.
“They have no control on the kernel or the OS.”
s/no/less/g
“Only companies that don’t have an OS dept like Dell or IBM’s PC division will opt for something like Linux.”
Therefore, SGI, HP, SUN, SCO aren’t companies…
Wrong again. Sun machines are being bought at larger numbers than before. Notice IDC reports on how sun’s server volumes have increased 26% year on year.
Sorry, you don’t listen to surveys, and certainly not those by people like IDC. Sun has sold enough when it breaks-even. It’s that simple.
Sun has SPARC/Intel servers from $2000 and up. How much cheaper did you expect?
Certainly, an x86-based server should start way below $2000. You can certainly get a pretty good AMD-based server for much less than that elsewhere, which makes a difference at the cheaper end of the market. Sun does not know how to sell here, and everything seems to be a sweetener to get you to buy SPARC and Solaris. People there don’t want it.
You’re right in a way though. Sun has had a lot of good kit, even at the low end, and I have seen an excellent Sun Cobalt Cube up and running for slightly less than $2000. The problem is the marketing, the message they are delivering and the fact that they have discontinued a lot of very promising product lines. The Cobalt Cubes were excellent low-cost servers but Sun bought them, and then decided they weren’t going to sell them any longer.
When will Sun understand that many people are not going to give up and buy more expensive SPARC servers running Solaris unless they actually need them? That’s why people are passing Sun over, because that is all Sun seems to want you to do.
Sun seem to have very little idea about what they do sell. A lot of what they sell, including desktops, is sold and supported very badly and the products are bad themselves. A lot of the time they don’t even seem to realize what is required technically – Network Computers or Java Stations anyone? The stuff that is good Sun does not seem to realize is good, and they start distracting people by giving out confusing messages or discontinuing very good product lines like the Cubes. That’s called slashing your own throat.
From your posts here it looks like all you know about stuff is the FUD linux zealots spread, go and do some real research before you post.
Ha, ha, ha. Yer. Sun’s bottom line and total lack of profitability is just FUD invented by Linux zealots
. Please.
Sun have a lot of good aspects, skills, expertise and products (if they don’t discontinue, continue and discontinue as they usually do). Unfortunately, the message that does not seem to have got through to people is that SUN IS GOING BUST. Investors are not interested in whether server shipments are up in someone else’s survey. They want to know what the situation is on the accounting documents, and it probably looks even worse inside than outside.
It is no good telling people about how cheap Sun’s servers are, and how well they are doing because they are going bust. That’s right – SUN ARE GOING OUT OF BUSINESS if their current direction continues. No matter how good and cheap they are they must be doing something wrong.
How many times do I have to point this out to clueless people who have no idea what being in the red actually means? Zealots or not zealots – Sun is in a very serious position.
Can LTC add any IBM centric mods to LTC and have Linus accept that?. Not in a million years!. LTC cannot use any of IBM’s patents and add that technology to Linux. Last time IBM tried that, they invited SCO’s lawsuit. Effectively LTC is a bug hunting division at IBM.
> “Only companies that don’t have an OS dept like Dell or IBM’s PC division will opt for something like Linux.”
>
> Therefore, SGI, HP, SUN, SCO aren’t companies…
Is English not your first language?.
I’ll rephrase the statement: Companies that don’t have their own OS division will use Linux. Eg: Dell, IBM PC, HP PC div,
Gateway as opposed to Apple or Sun or SGI’s MIPS/IRIX or IBM’s AIX/RS6K or HP’s PARISC/HPUX/ or HP’s Alpha/Tru64 will always rely on their inhouse technologies first.
Who said anything about SGI/HP/SUN/SCO not being companies???
“an LTC add any IBM centric mods to LTC and have Linus accept that?. Not in a million years!. LTC cannot use any of IBM’s patents and add that technology to Linux.”
xfs is one such ibm centric mod. numa related patents from ibm has been licensed for gpl use. so both of your claims are wrong. IBM HAS OS DEPTS. get that ?
“Who said anything about SGI/HP/SUN/SCO not being companies???”
That’s what i concluded from your statement. I thought it was so obviously bull that i didn’t need to argument anything except stating the obvious.
“Companies that don’t have their own OS division will use Linux. Eg: Dell, IBM PC, HP PC div
[…]
HP’s PARISC/HPUX/ or HP’s Alpha/Tru64”
It did not matter you rephrased it, it still makes no sense to me. You name HP and IBM in both rows, and Dell doesn’t have an OS division while SCO, SGI, SUN, etc do while they (refers to companies i named in this sentence) all are “using” it.
I think it is a load of hard-worded generalisation and i’ll give you 2 examples to think about:
NewSCO (Caldera) developed OpenServer and UnixWare while they also developed on Linux. I’d say that back with Ransom Love before the litigation garbage, NewSCO (Caldera) had Linux (their Linux distribution) as their core business as OS (x86) and not OpenServer (x86) or UnixWare (x86) anymore.
Another example is SGI has both stopped supporting IRIX on their older models _and_ helped/supported Linux on (some of these) older models (R4x00, R5000, and R10000 on the way). Ofcourse i do not know why SGI stopped supporting the older models, so it is hard to conclude, however it is obvious that your logic doesn’t apply here.
“xfs is one such ibm centric mod.”
XFS is from SGI. JFS is from IBM. Both are GPLed in the Linux kernel. One of the many contributions to the Linux kernel from IBM, SGI.
[I don’t see much contributions from SUN]
“numa related patents from ibm has been licensed for gpl use.”
NUMA is a SGI invention and indeed is supported by the Linux kernel and/or gpl patches (ie. on Altix).
“so both of your claims are wrong. IBM HAS OS DEPTS. get that ?”
Indeed both SGI and IBM have OS depts.
Your statement counts for both SGI and IBM though you had some things assigned on the other company.
“Last time IBM tried that, they invited SCO’s lawsuit.”
bull shit. ibm can use whatever patents it has for whatever purpose it needs. sco’s case is about contract obligations over a project to port unix to 64 bit architecture along with ibm which ibm cancelled midway
Yes. I confused jfs with xfs but ibm has numa related patents
lets see
sgi-xfs and numa
ibm -jfs and copy on write
fine
IBM numa related patents
http://216.239.51.104/search?q=cache:fcfw15jmYgIJ:www.research.ibm….
For example, IBM has several patents (including patent number 4,656,583) on workmanlike, albeit complex, speedups for well-known computations performed by optimizing compilers, such as register coloring and computing the available expressions.
Patents are also granted on combinations of techniques that are already widely used. One example is IBM patent 4,742,450, which covers “shared copy-on-write segments.” This technique allows several programs to share the same piece of memory that represents information in a file; if any program writes a page in the file, that page is replaced by a copy in all of the programs, which continue to share that page with each other but no longer share with the file.
——————
so ibm can and does license patent under gpl. it has several OS related depts
where do you think it develops aix and linux related stuff?. in a candy store?
You just made a fool of yourself. The fact is that nobody wants to maintain their own proprietary Unix anymore, except SUN. It’s a headache to have to do your own maintenance.
Linux is the default unix implementation. That is what is driving IBM and HP to support Linux. They are quite eager to not have to keep their own huge development staff and to be able to leverage their expertise in more specialized areas other than OS development. By the end of this decade, Linux will be the primary Unix sold both by HP and IBM. Sun, well, it remains to be seen whether it will be around by them. And Dell, will in two-three year’s time make a huge surprise announcement and begin selling Linux on all of their desktops.
Exciting times ahead…
> bull shit. ibm can use whatever patents it has for whatever purpose it needs.
BS again, if IBM wants to use a “patent” in the Linux kernel, it has to reliquish the patent because Linux kernel cannot contain any patentable code (read GPL). If IBM reliquishes their patent then every body else who has licensed or will license their patent gets a free ride – IBM’s not that stupid.
IBM is certainly at liberty to do what it chooses with their patent portfolio except that in 4 years of committing 1Billion to Linux they have yet to assign a patent over to GPL. So the likelyhood of IBM being benevolent is slim (not that it cannot happen).
> You just made a fool of yourself. The fact is that nobody wants to maintain their own proprietary Unix anymore, except SUN. It’s a headache to have to do your own maintenance.
Oh please don’t show your freaking ignorance: Apple has their own UNIX, Sun has their own UNIX SCO still does their own UNIX, WindRiver does their own Realtime UNIX, IBM still invests a bit into AIX, so does HP and SGI. Wasabi does NetBSD, and there’s still others doing their own non-linux operating systems – aka SkyOS. Nokia uses their own OS (maybe not UNIX but it’s still an OS).
BS again, if IBM wants to use a “patent” in the Linux kernel, it has to reliquish the patent because Linux kernel cannot contain any patentable code (read GPL).
Well i thought the GPLv2 doesn’t address this while the FSF does want GPLv3 to address this. Perhaps you can quote the specific lines in the GPL which make you believe quoted statement?
However i think you make a mistake. I think you do not know that one can say “Ok, X may use <patented thingy> for free (as in beer)” and say also “No, Y has to pay <insert # of money> to use <same patented thingy>”.
SW,
Go back to Fiorina’s and Palmisano’s statements during LinuxWorld and you will realize that the era of “not-invented-here”, so let us spend tons of bucks in our proprietary crap is over. Customers don’t want proprietary lock-in and the fewer customers you have the less profitable it becomes to keep your own development tree.
There is a diminishing return. I find it hilarious that people feel a need to bash an OS that they are free to tinker with, share and learn more about. An OS that has proved itself robust and that excels as both a server and a desktop, every day more so.
Ps: Attack the ideas, not the person. It’s the basis of debating 101
”
BS again, if IBM wants to use a “patent” in the Linux kernel, it has to reliquish the patent because Linux kernel cannot contain any patentable code (read GPL). If IBM reliquishes their patent then every body else who has licensed or will license their patent gets a free ride – IBM’s not that stupid. ”
no. there is no such thing in the gpl. gpl can very well contain patents licensed for use under it. if its licensed for use under gpl it can be used only for gpl code and everyone who modifies and distributes such code have to share it with others. nobody can get a free ride.
http://www.gnu.org/copyleft/gpl.html
read it and quote the relevant part if you can contest
IBM is not the only one who has such patents. recently pramfs was licensed for gpl use while retaining full commercial rights for any other purposes.
Read
http://kerneltraffic.org/kernel-traffic/kt20040418_258.html#2
Dont spread misinformation about gpl.
”
IBM is certainly at liberty to do what it chooses with their patent portfolio except that in 4 years of committing 1Billion to Linux they have yet to assign a patent over to GPL. So the likelyhood of IBM being benevolent is slim (not that it cannot happen”
you are completely ignorant
http://oss.software.ibm.com/developer/opensource/linux/papers/gpl.p…
IBM has patents related to copy on write. specifically IBM patent 4,742,450, which was licensed under gpl. we have several such patents. do your research before you post nonsense
“BS again, if IBM wants to use a “patent” in the Linux kernel, it has to reliquish the patent because Linux kernel cannot contain any patentable code (read GPL).
Well i thought the GPLv2 doesn’t address this while the FSF does want GPLv3 to address this. Perhaps you can quote the specific lines in the GPL which make you believe quoted statement? ”
gpl v3 doesnt address this because it isnt a problem. nobody needs to relinquish their patent. they can specifically license it under gpl and use it for other commerical purposes in whatever way they want. gpl v3 if it is designed addresses a different problem of patents contesting implentation itself. for example compatibility with apache license
“gpl v3 doesnt address this because it isnt a problem. nobody needs to relinquish their patent. they can specifically license it under gpl and use it for other commerical purposes in whatever way they want. gpl v3 if it is designed addresses a different problem of patents contesting implentation itself. for example compatibility with apache license”
It is what i claimed under GPLv2 however under GPLv3 i do not understand this. What exactly does GPLv3 according to you?
Why does my FSF t-shirt containing information of the GPL (i think GPLv3) contain the following wording
“[…]
Finally, any free program is threatened constantly by software patents. We wish to avoid the danger that redistributors of a free program will individually obtain patent licenses, in effect making the program proprietary. To prevent this, we have made it clear that any patent must be licensed for everyone’s free use or not licensed at all.
[…]”
Does the FSF mean “everyone’s free use” as in under the scope of the GPLed version of the software or does the FSF mean “everyone’s free use” as in every piece of software which implements the patent in question?
”
Why does my FSF t-shirt containing information of the GPL (i think GPLv3) contain the following wording
“[…]
Finally, any free program is threatened constantly by software patents. We wish to avoid the danger that redistributors of a free program will individually obtain patent licenses, in effect making the program proprietary. To prevent this, we have made it clear that any patent must be licensed for everyone’s free use or not licensed at all. ”
gpl v3 does not exist yet and hence you cannot have it in the tshirt. whatever your tshirt says isnt a license and hence does not have any legal value at all.
gpl v2 doesnt have any such patent obligations. patents can be specifically licensed for gpl use and has been done several times by ibm and others. however software patents are considered a threat by fsf and hence it will work towards eliminating it. no license can invalidate patents.
so called intellectual property stuff includes several different things like
copyright
trademarks
patents
they do not invalidate each other. gpl can contain patented stuff if it is compatible. trademark does not apply to gpl at all.redhat protects its trademark while licensing software under gpl for example
http://www.redhat.com/about/corporate/trademark/
I hate to state the obvious, but if you choose to use commercial software on Linux then you will probably be subjected to vendor lock-in. Most of the commercial software is only supported on RedHat or SuSE, this means that you are either not free to use the distro of your choice, or you have to run your software unsupported. The same thing goes for a lot of hardware, the vendor will only help you if you are running RedHat or SuSE. So vendor lock-in is very much alive and well in the Linux world, at least as we know it today.
“I hate to state the obvious, but if you choose to use commercial software on Linux then you will probably be subjected to vendor lock-in. Most of the commercial software is only supported on RedHat or SuSE,”
you are confusing commercial and non free
http://www.gnu.org/philosophy/words-to-avoid.html
i acknowlodge everything you said however you have not answered my question concerning GPLv3, Apache, and patents. I understand we can only speculate on it, but there must be some information on it, right? TIA.
I am not confusing the issue, if you use commercial software, “A program is commercial if it is developed as a business activity.”, and the developer will only support you if you use the software on RedHat or SuSE then you are essentially being subjected to vendor lock-in. The developer could have that requirement in place for a multitude or reasons, one being there are just to many ways to make a Linux install look and they only want to support it on a couple distros.
“I am not confusing the issue, if you use commercial software, “A program is commercial if it is developed as a business activity.”, and the developer will only support you if you use the software on RedHat or SuSE then you are essentially being subjected to vendor lock-in.”
no. commerical entities can only support a limited subset. if this is lock in then every commercial entity is a lock in.this is not what is referred to as lock in usually. you can always get support from a different entity when it comes to free software. however non free software can only have limited support (ie) hand holding.
“GPLv3, Apache, and patents. I understand we can only speculate on it, but there must be some information on it, right? TIA.”
there are several speculations
apache v2.0 is said to incompatible with gpl according to fsf
http://www.gnu.org/philosophy/license-list.html#GPLIncompatibleLice…
“3. Grant of Patent License. Subject to the terms and conditions of this License, each Contributor hereby grants to You a perpetual, worldwide, non-exclusive, no-charge, royalty-free, irrevocable (except as stated in this section) patent license to make, have made, use, offer to sell, sell, import, and otherwise transfer the Work, where such license applies only to those patent claims licensable by such Contributor that are necessarily infringed by their Contribution(s) alone or by combination of their Contribution(s) with the Work to which such Contribution(s) was submitted. If You institute patent litigation against any entity (including a cross-claim or counterclaim in a lawsuit) alleging that the Work or a Contribution incorporated within the Work constitutes direct or contributory patent infringement, then any patent licenses granted to You under this License for that Work shall terminate as of the date such litigation is filed.”
——————————————
this patent restriction is acceptable to fsf but it still is a restriction that is NOT in the gpl and hence said to be incompatible. however apache foundation doesnt agree
http://www.apache.org/licenses/GPL-compatibility.html
gpl v3 is speculated to have something similar and there are also plans to see if gpl v3 can be modified to invoke it when used as a ASP style model over the internet.
All these is mere speculation and fsf hasnt come forward with a draft form or a time limit
you are completely ignorant
http://oss.software.ibm.com/developer/opensource/linux/papers/gpl.p…..
So show me what file in the Linux kernel documents all patents that have been assigned over to FSF/Linus?
IBM has patents related to copy on write. specifically IBM patent 4,742,450, which was licensed under gpl. we have several such patents. do your research before you post nonsense
Except that this is a submarine patent.
http://www.markshapiro.com/Issue22.p11-p19.html
Patents are also granted on combinations of techniques that are already widely used. One example is IBM patent 4,742,450, which covers “shared copy-on-write segments.” This technique allows several programs to share the same piece of memory to represent information in a file. If any program writes a memory page to a file, the contents of that page is copied to all programs. The memory is shared between programs, but no longer shared with the file. Shared segments and copy-on-write techniques have been used since the 1960’s. Nevertheless, the Patent Office thought that it merited a patent, which must now be taken into account by the developer of any new operating system.
This isn’t one of IBM’s “family jewels” patent. It’s right up there with Cursor X-OR patent. This patent should be challenged as vigorously as MS has challenged Eolas. The fact that IBM doesn’t go after anybody implementing this patent implies that it’s a non-enforceable patent.
One last thing, Sun was telling us two years ago we’d all be running on glorified VT100 terminals. At what point do these clowns lose all credibility?
I guess that is my point, if you are using free software then there is no vendor lock-in. You can run Solaris, Linux, *BSD, AIX, etc… If I can find a lot of the same apps at http://www.sunfreeware.com, http://www.debian.org, and http://www.freebsd.org, where is there any vendor lock-in?
This is the way it should be anyway, I should be able to run whatever software I want on whatever OS I want.
”
So show me what file in the Linux kernel documents all patents that have been assigned over to FSF/Linus?
”
you are again misunderstanding the issue. a patent need not assigned to a entity like fsf or a person like linus to be used under GPL. Patent doesnt work like copyright and hence no document under the kernel is necessary.
what is required is a patent grant which says that it can be used under GPL. Such a patent grant for GPL doesnt contradict under restrictions under any possible license that the patent grant entity may use. This doesnt revoke the patent as free for all.
any ibm contributions that have patents have specifically been under a grant that allows it be under GPL. you might want to search in the lkml for such patents.
“The fact that IBM doesn’t go after anybody implementing this patent implies that it’s a non-enforceable patent.”
Again only a trademark can be made non enforcable if not effectively contested. A submarine patent isnt automatically made void if non enforced. it can be enforced at anytime IBM wants. IBM may have choosen to use this patent for defense instead of litigation
:I guess that is my point, if you are using free software then there is no vendor lock-in. You can run Solaris, Linux, *BSD, AIX, etc… If I can find a lot of the same apps at http://www.sunfreeware.com, http://www.debian.org, and http://www.freebsd.org, where is there any vendor lock-in?
This is the way it should be anyway, I should be able to run whatever software I want on whatever OS I want.
:
this is ideal. that fact that free software can also be sold in a commercial manner makes you statement confusing. you should have used the term non free if you mean software that is not modifiable
If you are referring to the Sun Rays, I know of one business that only has Sun Rays on their desks, one place that has a fare number installed and we are going to be piloting them this summer. The whole concept just takes a change in thinking. That change being that you don’t need to have a computer at your desk with an OS and lots of software. This is compelling for businesses looking to lower the expense of maintaining thousands of PCs.
Obviously for personal use it is better to have the flexability and freedom of a PC.
Sorry if I was muddying the waters, but I was only trying to illustrate the point that some software will “lock you in”, but just because you might be using an OS that isn’t open source doesn’t mean that you are locked in. I like to use Debian, but sometimes when I am looking at software, Debian is not supported, but RedHat and SuSE are. Then you have software like Apache, that can be run on a lot of OSes.
So if I choose to run Apache on Solaris, I would not be locked-in, I could simply move to freebsd or debian if I no longer wanted to use Solaris.
“Sorry if I was muddying the waters, but I was only trying to illustrate the point that some software will “lock you in”, but just because you might be using an OS that isn’t open source doesn’t mean that you are locked in. ”
depends on the software itself not the OS. in your example apache is OSS so you are free from lock in. so specifically it depends on the license of the software and the use of well documented open protocols instead of proprietary patented stuff that doesnt allow interoperability
I agree with you.
The only reason for my original post was that people think that somehow if you use a closed source OS that you are automatically some how locked into something. I would say that there is a greater chance for lock-in if you use closed source applications that require you to run Windows or RedHat/SuSE, or any OS for that matter.
If nothing else, because of OSS you can feel safer about running an OS like SunOS or AIX because if you decide that you now want to run Linux or *BSD you can still use your existing hardware.
”
The only reason for my original post was that people think that somehow if you use a closed source OS that you are automatically some how locked into something”
thats because it has traditionally been that way. for business needs it makes sense to encourage lock in. if you dont do that people will migrate easily if they think so.
in case of OSS people usually use open protocols. even when there is no documentation a competent person may derive it from the code itself. however with undocumented proprietary and patented stuff it is neither easy nor legal where the patent applies and the risk of lock in is much higher. this doesnt automatically imply that proprietary means lock in but the chances are pretty high
I think you and I are saying the same thing, it is just when I am saying it I am emphasising that it is still ok to use Solaris, or whatever OS trips your trigger for that matter.
I personally like the freedom of choice, I have been posting here on OSX. I could have used one of my other systems here in my cube running, FreeBSD, JDS, Debian, RHEL3WS, Solaris 9 Sparc, Solaris 9 x86, Win XP, or Win2k, the Mac just happened to be the one I went to osnews with first.
Wasn’t the topic about Sun’s pricing models?
who cares about GPL anyway… /. –> way
“who cares about GPL anyway… /. –> way”
a lot of people do. i could ask the same question about solaris. the discussion evolved into gpl and patents. if you consider that to be a problem use the report abuse button and see if the moderator agrees.
“gpl v3 is speculated to have something similar and there are also plans to see if gpl v3 can be modified to invoke it when used as a ASP style model over the internet.”
I’m too lazy too proof with sources, but this Apache license is new (january 2004?) while i’ve heard about the GPL patent changes far earlier (september 2003?). Not sure on dates, i’d have to look them up to verify, but i don’t have time currently. Maybe something to consider…?
@ the person who replied to chello.se troll: me suggests you don’t feed the trolls. The chello.se person is a well known anti-linux troll. Just ignore it, it ain’t worth it.
”
I’m too lazy too proof with sources, but this Apache license is new (january 2004?) while i’ve heard about the GPL patent changes far earlier (september 2003?). Not sure on dates, i’d have to look them up to verify, but i don’t have time currently. Maybe something to consider…? ”
why speculate on this too much. there is been a lot of dicussions. we can consider this if and when the license is posted
“why speculate on this too much. there is been a lot of dicussions. we can consider this if and when the license is posted”
Because i like to speculate based on hard facts so i am able to foresee the future. It is my spice. Later.
Let me try again.
You’re an illiterate moron, buddy. What the hell did they teach you at school? CPU architecture itself can not be automatically classified to some memory access class (NUMA, SMP, etc.). The same CPU architecture can theoretically be used in any architectural class.
It has an onbard memory controller. Any implementation will be NUMA. If that is too much for your brain to handle, just think about current Opteron NUMA architectures.
> Oh, and Sun’s latency figures are actually pretty comparable to the Altix for your information.
Memory latency is not a real issue here.
Yes it is. Primarily latency, secondly bandwidth.
Enterprise applications (databases especially) are primarily spaghetti code with a lot of branching and therefore requires a significant degree of cache locality for optimum performance. Accessing remote memory is very expesive in NUMA and CC-NUMA in particular,
Do you even know what the difference between NUMA and ccNUMA is? I suspect not. Anyway, *why* do you think accessing remote memory is expensive, Einstein? Duh maybe because of latency?
thus NUMA will be considerably less efficient than SMP working on the same workload executing a tipical enterprise application — the performance of CC-NUMA machines is severely limited if data is incorrectly placed at physical memory locations away from the nodes that use it most, if data needs to migrate from one set of nodes to another, or if an application requires a large amount of data to be replicated on each processor.
With all else being the same, of course the system with slower access to some parts of memory will be slower.
However, SMP systems just can’t scale up to as many CPUs as NUMA architectures, nor will they be as efficient. I suggest you take a course in computer architectures if you don’t understand that.
I suggest that you stop trying to pick trivial “holes” in my arguments (ie. Opteron not NUMA), because you were wrong even on that point. And you are wrong on the main point of the argument, which is that Sun’s systems *are* NUMA. Had you conveniently forgotten that? And you don’t even seem to understand how NUMA works or basic concepts like the memory heirachy.
Just admit you’re wrong. Trying to argue with me is just making you look like more of an idiot because you obviously don’t know anything.
Sorry, you don’t listen to surveys, and certainly not those by people like IDC. Sun has sold enough when it breaks-even. It’s that simple.
Or if it can reduce it’s expenses to cover for the loss in margins, basics of business, it’s that simple. Duh. Bottom line is Sun’s volume is larger than before, contrarty to your “not selling enough”. More customers are buying Sun boxes and in larger volumes than before but at lesser prices than before (sun is selling more lower-end, lower-margin systems). Profits are a function of revenue and expense, Sun can reduce expenses and still make profits at the current volumes. And they are doing just that, the recent job and product cuts are an example of that.
IDC is a well trusted source on server sales and volumes, the whole industry depends on thier surveys.
Certainly, an x86-based server should start way below $2000. You can certainly get a pretty good AMD-based server for much less than that elsewhere, which makes a difference at the cheaper end of the market. Sun does not know how to sell here, and everything seems to be a sweetener to get you to buy SPARC and Solaris. People there don’t want it. [i]
Sun’s x86 line is competitively priced with it’s market compertitors IBM, HP, and Dell. In terms of selling a server, OS and support, the total cost is very competitive. Care to provide a link to a comparable system from a vendor for “way below” $2000 as Sun’s lowest end xeon/opteron box?
[i]Sun seem to have very little idea about what they do sell. A lot of what they sell, including desktops, is sold and supported very badly and the products are bad themselves. A lot of the time they don’t even seem to realize what is required technically – Network Computers or Java Stations anyone? The stuff that is good Sun does not seem to realize is good, and they start distracting people by giving out confusing messages or discontinuing very good product lines like the Cubes. That’s called slashing your own throat.
The internet Appliances market died with the dot com boom. So It looks like your perspective of what a good product is differs from everyone else.
Ha, ha, ha. Yer. Sun’s bottom line and total lack of profitability is just FUD invented by Linux zealots
. Please.
No the ignorant talk about Sun’s products and market.
It is no good telling people about how cheap Sun’s servers are, and how well they are doing because they are going bust. That’s right – SUN ARE GOING OUT OF BUSINESS if their current direction continues. No matter how good and cheap they are they must be doing something wrong.
More FUD. Sun isn’t going out of business anytime soon, Sun just announced key products, alliances and cost reduction to help them get out of the re.
How many times do I have to point this out to clueless people who have no idea what being in the red actually means? Zealots or not zealots – Sun is in a very serious position.
Being in the red means what I pointed out above. You are the clueless one to think that the only way to be in the red is to not sell enough. You can sell all you want and have expenses much more than your revenue and still be in the red.
Notice IDC reports on how sun’s server volumes have increased 26% year on year.
I’d be curious to know what was the increase of Solaris servers, though.
For 2003, IDC reported a 35% percent increase in the sale of Linux servers overall – and Sun was definitely not in first place (Dell had the highest increase, IIRC).
Meanwhile…the anti-Linux trolls are coming out in numbers today, aren’t they?