Slashdot Mirror
Sun Unveils More Linux Strategies
A number of people have submitted the press release from Sun Microsystems about their latest announcements in conjunction with Linux. Highlights from this one include the promised release of "New single- and multiprocessor systems, to be announced mid-year, will use the x86 architecture and be capable of running thousands of Linux applications natively." As well, they are expanding the Cobalt line of servers, but even more interestingly they are going to "freely offer" parts of Solaris - but no license specified that I saw. They are also releasing "ABICheck", which should check compatibility between Linux/Solaris. C|Net is carrying coverage now as well. And it looks like Lineo and SuSe are going to get competition in the embedded and telecom support area - I wonder if that's tied to the OSDL announcement. It's good to see that they are getting on the right track - now let's hope they stay the course.
Can now go and retract all the Sun naysaying.
I use Solaris for SPARC, its great, but Solaris X86 was half-baked from the start. The writing was on the wall for a LONG time, but when Sun finally canned it, I for one had to endure both the cries of "abandonware!" as well as generic sun bashing from the local Linux people I have to deal with.
It should be obvious now, Sun is doing the right thing by ceeding the X86 market to Linux, and infact helping the transition, for those that were in the Solaris X86 crowd. Win-win situation, as far as I can see.
Anyone who considers arithmetical methods of producing random numbers is, of course, in a state of sin.-John von Neumann
i can not believe that sun has let cobalt stagnate to the extent that they have. i very much like the management capabilities of the machines, but i would really like more resources for experimenting. the party line is that if you mess with the system it is unsupported. quite sad that they are pretty much where they were two years ago--k-6 in the raq 4's!
> I understand the BSD structure resembles SunOS and HP/UX more than Linux
:)
Not really, SunOS used to be BSD based but changed that alot when they combined their stuff with AT&T's and ended up with System5Release4. Depending on the Linux distro sometimes Linux will end up closer to a SVR4 Unix.
> I keep wondering why big companies like HP and Sun choose linux, instead of freeBSD
Guess what OSX is based on
My Hello World is 512 bytes. But it's also a valid Fat12 boot sector, Fat12 file reader, and Pmode routine.
bandwagon and mindshare... CIOs have heard of Linux, not all of them have heard of *BSD. That's sad, because the BSDs are far more mature at a system level and I think they probably scale better. Then again, Sun and HP have Solaris and HP-UX for selling scalability.
Something else they're not thinking about is that Linux is not 100% POSIX-compliant. that's going to piss off a lot of senior engineers who have to port legacy apps from HP-UX/Solaris (or, shudder, older Unixes) over to Linux.
Does this mean they are going to be nicer to the Jakarta folks???
http://jakarta.apache.org/site/news.html under "30 January 2002 - That flaming fireball in the sky..."
Sun's always been friendly to OSS as long as it gives them good press to be so. I'm not certain they are so good at heart. Maybe they were just scarred by microsoft changing the meaning of Java that they don't trust an ad-hoc group of unpaid developers to not do the same.
Karma Clown
I just cannot see Sun replacing Solaris on their high-end multi-processor machines... Or at least not until Linux scales equally well :-)
Because Linux has mindshare and is a big buzzword at the moment. Make no mistake, the big guys are behind Linux because it's a good marketing move. I applaud the efforts of IBM, HP, and Sun in this regard - but one should be very careful before getting into bed with them.
Linux is a convenient tool for IBM to rescue their big iron from obscurity, for HP to save themselves from obscurity, and for Sun to sound like they're not falling behind IBM and HP.
There's an argument that picking a GPLed OS means that competitors can't commercialise their work, but I'm not convinced about this one. If you look closely, few lines of code have come out of these houses. They're much more interested in making sure that their hardware can run Linux, or Linux apps, than in supporting the general Open Source / Free Software movement. It's a careful play to ensure that if the OS ends up being commoditized, people don't pick Intel's hardware.
I keep wondering why big companies like HP and Sun choose linux, instead of freeBSD.
I can't speak about Sun or HP, but some time ago SGI started working on tons of stuff for Linux, including but not limited to their XFS filesystem. More info: http://oss.sgi.com.
It's pretty clear, when you think about it, why they chose to release their valuable technologies for Linux rather than BSD: the GPL. GPL is, contrary to what Microsoft might say, a pretty business-friendly license. If a business spends billions of dollars over decades developing, say, XFS, then releases it under a BSD-style license, then anybody can incorporate that technology into their commercial products for free.
On the other hand, releasing XFS for Linux under the GPL means SGI gets to say they have XFS on IRIX and also on Linux, but it does not mean that Sun can put XFS in Solaris or whatever.
You can't make any money, directly, off of producing GPL'd code, but you can at least prevent your competitors from benefiting from your work.
I keep wondering why big companies like HP and Sun choose linux, instead of freeBSD. Although I'm not an expert on any of them, as far as I understand the BSD structure resembles SunOS and HP/UX more than Linux. Both BSD and linux are open source, and the BSD license even seems to be preferable to companies if, in the end, they decide to go closed source anyway.
Can someone explain this to me?
Because developers tend to prefer the GPL, which garauntees that we won't be buying back the fruits of our own labor one day from someone who's taken the whole thing, added some decoration, and used it to are part of some kind of toll booth on the information superhighway.
Life's a bitch but somebody's gotta do it.
Maybe Sun will make a new x86 system that has improved I/O -- like, using UPA rather than (or in addition to) PCI.
Since Sun will not be worrying about Windows support, they can extend the architecture a bit. Still use x86 processors, but enhance the surrounding systems to make it less PC-like and more big-server-like.
Napster-to-go says "Fill and refill your compatible MP3 player", which is a lie. It's not MP3. It's WMA with DRM.
Is their work in Linux part of a long-range strategy to phase out Solaris?
Unless it is done right, Linux on really big servers won't be quite as good as Solaris. Sun has invested a lot of effort in making Solaris extremely efficient on many processors. Sun can afford to drop Solaris only if Linux is equally good or better on large computers, which isn't the case, right now.
Instead, Sun sees Linux as an opportunity to position themselves better against small-time servers, such as those that run Windows NT/2000.
Healthcare article at Kuro5hin
IBM and Compaq find it pretty lucrative as well as Sun.
And I'm sure I could find dozens upon dozens of failed low-end hardware providers too. Infact, if you have this innate sense for being right, why arent you a rich man already, seeing as how you obviously know in advance who is going to succeed in a market, and who isnt, in advance.
Sun's main purpose in life seems to be as the launching platform for Oracle. Some of Sun's competitors have better performance, some have better prices, some claim to have both, but nobody has the level of Oracle support that Sun/Solaris gets. Without Oracle, there would be no Sun. Considering Larry's announcement about migrating all of Oracle's corporate systems to Linux, the handwriting is on the wall for Solaris. From Oracle's perspective, Linux is a great way to enhance their position vs. M$ SQL server on the low end, and go after IBM DB2 on the high end, all at the same time.
If anyone believes what Larry says, it looks like Oracle will elevate Linux to the top tier of supported OS, probably at the expense of Solaris. This really sucks for me because I committed to the SPARC/Solaris platform about 8 months ago. Oracle support of Linux wasn't quite there yet and I didn't have time on my side. I always thought a transition to Linux was inevitable, but I thought it would take another year or two.
From Sun's point of view, they are probably looking for a smooth way to transition SPARC Solaris to SPARC Linux, so as to drop Solaris entirely as a cost-cutting measure. Sun needs either a huge boost in SPARC CPU performance or lower pricing, preferrably both. Otherwise they will get killed by high-end X86 systems.
I think the ultimate fate of Sun/Solaris will be the same as Digital/VMS: It's another attack from the commodity boxes, armed with a standard operating system, this time without the M$ nonsense.
If you can easily replace a low-end Sun system with an X86 system, then do it. You werent a lucrative customer of Sun anyway. Sun makes their money off the kinds of systems that you cannot replace with cheap X86 hardware anyway. Same with IBM, its just they have an option for Linux instead of AIX.
So, asthetics aside, and knowing that its the hardware sales that drive profit, what is the difference between IBM and Sun? To the point that IBM are seen around here to be so wonderful, and Sun to be an evil corporation with declining marketshare, and whatever else people want to prophesize to the detriment of Sun?
Not much? Thought so. We now return you to the normal Slashdot blinkered advocacy...
Slashdotters are so ignorant of the various classes of machines and thier specialized purposes. Cray's were very specialized towards one particular type of computing that the government wanted. Cray may be dead, but supercomputers aren't. And anyway, Cray technology lives on in Sun's E15000.
Compaq just built the Pittsburgh Supercomputer Center's system.
Compaq and IBM are making Supercomputers for Department of Energy (read big nuclear simulations and whatnot).
And anyway that's not the high end that is important. Try running a really big data mining application running oracle on 32-bit x86 based Linux. It won't get very far. Could Celera have even begun processing the data for the Human Genome Project without 64-bit Alphas? No Way!
Just because you get a shell prompt on your beige-box, don't assume it can do everything that a real Unix machine can.
um SGI died because it tried to make windows a high end solution, unti they realized that was a dumb idea. besides sgi was more a workstation vendor.
also HP used to be an awesome high end vendor, but they made the same mistake as sgi. now they are pretty much another shitty pc manufacturer getting their ass kicked by dell.
"few lines of code"?????
HP is actively involved with Samba and with open source printer drivers. IBM is actively involved with JFS just like SGI is with XFS. All of them have contributed various bits of code or patches to the kernel at one time or another.
Do some research.
Linux fan that I am, it's nowhere near ready for the Starcat (E15K). That beast will take 72 processors (106 if you _really_ want them). Solaris has been doing large numbers of processors for a while (E10K, the 24 processor SunFire range, etc). It's pretty good at using them effectively. The linux kernel isn't there yet, in fact I think 8 processors is pretty much its limit right now......
..
The main factor limiting getting past this limit is that few people have access to this sort of hardware to do the development work. Think of how Linux got on the mainframe, a few bored IBM engineers had an old mainframe and got hacking. It got into the wild mostly because that implementation runs on top of the normal mainframe OS, and can co-exist with other mainframe apps. It got into production mainframes for precisely that reason. If it had required the mainframe to be dedicated to Linux we'd still be waiting.
Can the same happen with the E15K's? I don't think it will. Why? Because you'd have to run the Linux kernel on top of Solaris! This simply doesn't make sense when you'd be far better off running the apps natively under Solaris. The only way I see Linux getting onto that sort of hardware is if Sun (or IBM ) give access to one of these multi-processor machines to some developers. That's the short-term view. The linux kernel will continue to scale better and better, and I have no doubt it will get there, but for Sun to have mentioned it in that press release it would have to be there now, and it obviously isn't.
Besides which, you try convincing a conservative IT manager to spend US$1M+ on an E15K to run Linux on it, when you don't have successful case-studies to show him.
These views are not endorsed by my employer, and are given solely on the basis of public-domain knowledge, so don't try reading too much into them.
Stealing a rhinoceros should not be attempted lightly.