The $11.6-billion Sun Microsystems Inc, will soon unveil a brand new subscription-based pricing model for its Solaris 10 operating system (OS) in India. The Solaris 10 price plans will be determined based on the total number of CPUs (central processing units) at client locations and will be computed annually. Also, Solaris 10 ably equipped with fixes and features. Update: More on Solaris.
”
Solaris 10 will not tolerate Linux partitions on the same drive, so if you want to dual boot, you’ll need a separate hard drive. ”
that explains the whole Jonathan attitude
“that explains the whole Jonathan attitude”
Actually, from what I gathered from a blog somewhere, the reason Solaris 10 and Linux have problems is due to the ID numbers assigned to the partitions. There was something mentioned about how Linux likes to swap over the Solaris partition, which is not tasty, IMO. But I think that this has been fixed, however.
“There was something mentioned about how Linux likes to swap over the Solaris partition, which is not tasty, IMO”
whatever the reason, this is not going to happen. swap should be setup explicitly in both operating systems during installation
What about this: http://www.sun.com/2004-1130/feature/
It appears Solaris 10 is completely free for those who don’t want support, and, then, there is a tiered pricing scheme for increasing amounts of support.
Don’t forget, however, that there are mailing lists, forums, docs.sun.com, and sunsolve.sun.com for very basic support, so the free option isn’t totally support-free.
I heard Solaris Ten is supposed to be open source. How do I download the source code?
I read it will be available after Solaris 10 is officially released in late January.
Not available. all you have is press releases from Sun at this point
“whatever the reason, this is not going to happen. swap should be setup explicitly in both operating systems during installation”
It is. Only that Solaris x86 partition id is the same as the linux swap partition ID. The linux distro’s installer thinks it has more swap space and claims it. Thus destroying
Solaris’s partiton. There is hack to change the solaris partion ID to something else using fdisk and then run the linux installer.
I don’t think this has anything to do with Sun’s or Solaris’ attitude. It is a greedy linux installer issue.
Just another more of your misinformed FUD.
According to this thread
http://www.linuxquestions.org/questions/archive/20/2004/08/3/45622
all you have to do is use the Solaris boot loader, which will boot a Linux partition just fine. It’s Linux that doesn’t boot Solaris partitions properly.
“I don’t think this has anything to do with Sun’s or Solaris’ attitude. It is a greedy linux installer issue. ”
linux doesnt have a standard installer. so your whole issue is bogus.
“It’s Linux that doesn’t boot Solaris partitions properly.”
you are probably talkign about grub or lilo. again Linux doesnt have its own boot loader
thats a thread on Solaris 9 and LILO both of which are history
The partition ID collision btw Linux and Solaris has been fixed in Solaris Express 11/04 aka beta 8 aka build 72.
The old identifier was 130 (0x82) and the new identifier is 191 (0xbf). The fdisk command allows switching the partition id back and forth from ‘Solaris’ (0x82) to ‘Solaris2’ (0xbf).
linux doesnt have a standard installer. so your whole issue is bogus.
May be that’s the problem. Linux needs more standards. I should have said Redhat and it’s dervivative distro’s.
May be Jem Matzan needs to clarify what he meant by Solaris 10 doesn’t tolerate linux. I think that statement is false.
You are basing your discussion on a false statement or I should say an ill phrased statement. The only issue with Solaris 10 and certain linux distros coexisting is the partion ID issue. And only certain linux distro installers clobber Soalris partions. So it is a linux not being tolerant of Soalris not the other way around.
The statement made by Jem is wrong on so many levels.
1) This is not a Soalris 10 issue. This was the case with Redhat 6.0 and Soalris 8 as well.
2) There is enough documentation online to prove that certain linux installers clobber Soalris partitions along with workaround.
I know becuase I dealt with this issue first hand in college while playing with dual booting a Solaris/linux box.
Please keep your misinformed opinions to yourself.
I find it funny that you took one statement from an entirely positive article, an incorrect one at that and are defending it. Especially when you don’t even know what the issue really is.
You totally missed this:
Many people will wonder, “Is Solaris 10 better than Red Hat Enterprise Server 3, Windows Server 2003, and SUSE Linux Enterprise Server 9?” Under most conditions the answer is yes, thanks to the above-mentioned features that are unique to Solaris 10. While SLES9 has Usermode Linux to do operating system virtualization, it requires assigned system resources and doesn’t offer optimal performance. Solaris Containers require only storage (hard drive) space to work and don’t suck up as much system resources, making this feature more efficient while providing similar functionality. ReiserFS v4 may be a significant step forward for Linux file systems, but looking through the feature list on its Web site, I don’t see anything like the ability to add storage space dynamically or integrated checksums to protect against data corruption. ReiserFS v4 is also not 128-bit, so its ceiling is much lower than that of ZFS. DTrace has no equivalent anywhere, as far as I can tell.
Since you brought those exact projects up in another Solaris article.
Sun has a linux Soalris dual boot how-to here:
http://wwws.sun.com/software/linux/docs/dual_boot.html
This has been discussed at length here:
http://www.linuxquestions.org/questions/archive/20/2004/08/3/45622
google for more.
May be that’s the problem. Linux needs more standards.
The problem is that while you have several individuals (users of proprietary software on Linux) and ISVs pushing for a standardized platform which will provide future binary compatibility for all applications, this attitude runs completely counter to the attitudes of the system developers, namely Linus on the kernel side and the GNU developers writing tools like glibc. They would rather Linux remain an open platform unfettered by past design mistakes, even at the cost of binary compatibility.
I’ve never seen the glibc developers consider the ramifications of changes they make which break binary compatibility. Often all they are able to cite are ambiguities in the POSIX specs which allow them to even make nonstandard changes as when they broke fgetpos()/fsetpos() by changing fpos_t into a structure, which also broke *source* compatibility of several applications which expect fpos_t to be an integer. No justification was given for why the change was made, simply that the POSIX specifications did not require that fpos_t be implemented as an integer (which is, if anything, an oversight in the specifications). The move had no justifyable technical superiority, it was nonstandard as compared to all other platforms in existence, and it broke binary compatibility. And GNU’s attitude is “Sorry folks, if you use Linux you just have to deal…”
An attitude of not allowing past design mistakes to hinder future development works great for a completely open source system, but is absolutely abysmal for proprietors of commercial software. About the only point binary compatibility can be found at is the system call interface, and even that changes from time to time.
This is a really bad position to be in. Because binary compatibility isn’t ensured, proprietary software is doomed to a future of binary incompatibility, all of the bugs contained within the statically linked application layer will remain until a future version of the software is released, and new features cannot be utilized.
Compare a proprietary application that is statically linked with LinuxThreads to a Solaris application which used Solaris 8’s previous M:N threading library. After a simple LiveUpgrade to Solaris 9, the Solaris application can take advantage of the new threading library. However, the Linux application is doomed to the poor performance of LinuxThreads until a new version is released by the vendor. If the application vendor chooses not to statically link, then you run into the problem that the application will break the next time you upgrade glibc.
Static linking is more of a band-aid for Linux’s lack of a standard ABI than a solution. I believe, especially with the advent of Open Solaris, that while Linux will remain the predominant player in completely open source systems, that Solaris will increasingly become an alternative to fill the role of an open POSIX platform for proprietors of commercial software. With Sun’s already massive vendor backing by such big name players as Oracle and the thousands of engineering and scientific applications already on Solaris/SPARC, I hope very soon we’ll see Solaris 10/IA32 and Solaris 10/AMD64 versions of many of these applications, and the fledgling role of Linux/IA32 as a low-end Oracle server completely snuffed out by Solaris 10.
So bottom line, Linux doesn’t need more standards… proprietors of commercial software need a better platform, thoroughly standardized with vendor support and ensured binary compatibility. Linux’s current model has been doing quite fine serving the needs of the open source community and the people at GNU seem to have little concern for making a good platform for proprietary software. This is unlikely to change in the near future, so I certainly hope ISVs begin looking to Solaris 10 instead.
In list form, for easy digestion
1) Solaris 10 looks very cool.
2) Sun’s engineers are awesome.
3) Sun’s management is clueless.
4) Not welcoming the competition is counter to the spirit of Linux.
5) When is ZFS gonna be in the Solaris Express builds?
Sun’s management is clueless.
If there’s one thing Sun needs, it’s a top notch PR department that is able to assess Sun’s industry image and work on ways of improving it. It seems a massive amount of their PR comes straight from the mouth of Schwartz, who manages to make Sun look confused and directionless when in reality it’s shaping up to look much better than it did a year ago when Sun’s doom seemed imminent.
As for their management being clueless, that’s not necessarily true. Schwartz was the man behind Project Mad Hatter and Project Orion, which became the Java Desktop System and Java Enterprise System respectively, both of which have been triumphant successes for Sun.
Schwartz knows what to do, just not what to say. Sun really needs to work on keeping their foot in Schwartz’s mouth.
When is ZFS gonna be in the Solaris Express builds?
I don’t think that’s going to happen. As far as I know, the general availability of ZFS will postdate the general availability of Solaris 10 itself.
”
Schwartz knows what to do, just not what to say. Sun really needs to work on keeping their foot in Schwartz’s mouth.
”
exactly….
”
I don’t think that’s going to happen. As far as I know, the general availability of ZFS will postdate the general availability of Solaris 10 itself.
”
not is project janus going to be there in the initial release
I can buy that. I don’t follow Sun’s management very closely, but every news story I see, it seems Schwartz’s position changes. First they have no “Linux strategy”, then they release Linux workstations, then they start criticizing RedHat. Indeed, the Solaris 10 launch seems to be their first concentrated and well-disciplined marketing effort in quite awhile.
May be that’s the problem. Linux needs more standards.
Relax. That was a sarcastic response to a stupid statement. Linux is fine the way it is.
Not available. all you have is press releases from Sun at this point
Gee, seeing as how Sun hasn’t shipped Solaris 10 yet, do ya think you could give them a break? All they’ve done is make preview builds available. Solaris 10 FCS won’t ship until January.
And Sun has just submitted their open source license to the OSI for approval. You can read it here – http://www.sun.com/cddl/
5) When is ZFS gonna be in the Solaris Express builds?
ZFS will not be in the initial build of Solaris 10. It will most likely ship in Update 1 (which should be a few months after 10 ships in January). I don’t know when it will appear in Solaris Express, though.
“Gee, seeing as how Sun hasn’t shipped Solaris 10 yet, do ya think you could give them a break? All they’ve done is make preview builds available. Solaris 10 FCS won’t ship until January.
”
you know thats actually a extremely short time to get it certified from OSI or get a community around it. all I hear is hype on unfinished stuff like project janus and ZFS. with people like confused and lier Jonathan leading this marketing effort its a total disaster. the technology is good. the top level people are not…
“And Sun has just submitted their open source license to the OSI for approval. You can read it here – http://www.sun.com/cddl/“
thats not true. SUN nowhere has claimed or admitted that this is the open source license for solaris nor has OSI certified it as open source.
in this case, standards really, really are needed. i’m assuming this whole problem occurs because way back when in history, linus, gnu, or whoever worked on linux fdisk program settled on a partition id without checking on other os’s that may use the same one.
.. and how would he check? afaik, there’s no official consortium or body that maintains fdisk partition ids. imagine if it was like this for mac addresses or anything else that requires *globablly* unique identification.
anyways, thank goodness it’s fixed now in the latest solaris express (build 72 i think).
Solaris 10 FCS wil ship in January. Sun hasn’t committed to a release date for OpenSolaris. They’ve only said that it will be under an OSI-approved licence. As for building a community around it, there are already lots of people eagerly awaiting release, especially the folks at Blastwvave. Yes, the OpenSolaris community will be small at first, but it will grow over time, as with any successful open-source project.
As for “only hearing hype,” why don’t you download Solaris Express and try out DTrace, Zones, FireEngine, and SMF?. Build 72 is quite nice (I already have it on several x86 and SPARC systems).
What basis do you have for calling Jonathan Schwartz a liar? It seems your reply is just more of the same Sun bashing that’s become fashionable. Since you can’t attack Sun’s tecnhnology, you have to attack the people behind it.
As for the CDDL, I never said it was OSI-approved, only that Sun had submitted it to the OSI for approval. And if you read carefully, I never said that CDDL was the OpenSolaris license, only that Sun had authored it (of course, if you put 2 and 2 together, it stands to reason that OpenSolaris will be licensed under CDDL).
Sergey: “The partition ID collision btw Linux and Solaris has been fixed in Solaris Express 11/04 aka beta 8 aka build 72.
The old identifier was 130 (0x82) and the new identifier is 191 (0xbf). The fdisk command allows switching the partition id back and forth from ‘Solaris’ (0x82) to ‘Solaris2’ (0xbf).”
Exactly. I have used Linux for years, and was horrified to find out that 82 was used for Linux swap and Solaris data. The obvious thing that came to mind was “Well, that is a mistake…I guess the Linux swap partition ID number will change soon.”. 7+ years have passed…maybe 10+.
Linux (kernel and kernel tools) do so many things right that this obvious goof just doesn’t make sense.
While the problem is clearly Sun’s, it is still the fault of the base Linux tools and kernel (not the installation tools) that this has dragged on for so many years. (I want to say 10+, though I don’t know when the partition numbers were last chosen for Linux.)
good thing openSolaris is coming, as I am too cheap to actually buy the commercial version.. plus we can to fix all the things eveyone complains about
eh, ill probably buy it but i reffuse to buy RHEL!!!
If you haven’t noticed Schwartz puts out a lot of statements out there to get public reaction. He isn’t clueless. McNeely, however, I think has been clueless at times but i think hes on track now
Wait…it’s not free as in speach and as in beer?
I’ve been lied to!
If you haven’t noticed Schwartz puts out a lot of statements out there to get public reaction. He isn’t clueless. McNeely, however, I think has been clueless at times but i think hes on track now
Well, with Schwartz, what he puts on his blog, is him thinking out aloud. If you don’t like reading what he is thinking out aloud, well, don’t read it
As for McNealy, for all the whining, he isn’t that bad; alot of what he says makes alot os sense; especially in regards to IBM and its lust to sell complicated setups with expensive services rammed ontop.
As for Microsoft, hell, who wouldn’t have a cheap shot; when I used to read the transcripts, it wasn’t a violent hatred of Microsoft, just a friendly shot over the good ship Microsofts bow.
This is a solaris thread. Why was linux brought up at all?
”
What basis do you have for calling Jonathan Schwartz a liar?”
because he lied about several things. last of which was calling redhat a proprietary OS and not LSB compliant
“because he lied about several things. last of which was calling redhat a proprietary OS and not LSB compliant”
Can you please post a link where he says that Redhat is not
LSB compliant. (this is not an assertation that such a
comment was not made, I would just like to see what he said).
Having the source available doesn’t necessarily mean that
something is not proprietary. Open standards and standards
compliance are not the same (or even genetically related) as
open source.
At best OpenSolaris will be as open as BSD and be about a popular as BSD. Even though it’s said (And may be true) that BSD is better then Linux, Linux has left BSD WAY behind. If not for Apple BSD would (Besides at hosting companies) be all but forgotten.
For one, no one is ever going to use Solaris for their everyday OS. (Not even developers) Maybe a few power users but that is about it.
Also so for a person who has been a Windows admin or user it is MUCH more easy to install and use Linux then Solaris. Even though Solaris is better etc once it’s up.
Also unless you install Gnome or Sun JDE on Solaris (If you can figure out how to do it) the gui emviorment SUCKS. (They still have CDE by default! Come on)
Solaris is an OS meant to be used by professionals that need to get a job done, not slack-jawed yokels with a spare PC to fuck around with. So I really don’t think they’re going to care that it’s hard for some end user to install. It’s not for them.
*Sigh* Here we go again.
Ok. Sun is already shipping JDS with Solaris 10. If you download the current Solaris Express build, it’s already there. How, exactly, is Solaris harder to setup than Linux? There certainly is room for Sun to improve the installer and the default config (like, not leaving so many network services on by default), but it’s really not that hard to configure the machine. I just installed Solaris Express build 72 on 3 machines yesterday, and I had the machines configured as I wanted within 30 minutes of the install.
And I’d hardly say that the *BSDs are “forgotten.” They don’t get anywhere near the media hooplah that Linux does, but there are still large, thriving developer and user communities around all of them.
good thing openSolaris is coming, as I am too cheap to actually buy the commercial version..
The commercial version will be free. You can’t afford free?
“Solaris is meant to be used by professionals. SGI caters to special effects companies and to governments. 3dLabs builds only high end video cards, ….”
The previous statements are correct. The only problem is that hardware vendors have shareholders who expect a profit at the end of the year. Whether the product is of professional grade or not doesn’t matter when it comes to the wallet.
The IT industry graveyard is filled with corporations that looked down on “slack-jawed yokels”. Anybody remembers Intergraph ? There is a lesson to be learned in the fact that Sun is selling opterons, chips designed by AMD, a company that was once synonymous of low quality computers. What about Intel ? They made a fortune selling chips (indirectly) to “good nothing overclockers” and other enthousiast home users.
In an cutthroat industry, is it really wise to be snob ?
“Solaris is an OS meant to be used by professionals that need to get a job done, not slack-jawed yokels with a spare PC to fuck around with. So I really don’t think they’re going to care that it’s hard for some end user to install. It’s not for them.”
And I guess that is why Sun has not made a profit in almost 3 years. Pros are REALLY snappin that Solaris up. Guess that is why they now have to give it away. And RedHat is making money selling Linux. (And making a profit!) And now Sun has to try and take the Linux approach by making Solaris open source. And you think Sun is not dead. How long do you think a company can survive without making a profit?? Can’t just keep sueing Microsoft to make money!
And I’d hardly say that the *BSDs are “forgotten.” They don’t get anywhere near the media hooplah that Linux does, but there are still large, thriving developer and user communities around all of them.
The only company making something out of BSD is Apple. This year alone a couple of BSD groups died. All that are left are Open BSD, Free BSD and Apple. Apple is the only one on that list that has a real product and that product is losing market share every day to Linux. Next time Bill Gates mentions that he is worried about BSD or Solaris the way that he is running wild about Linux you call me.
Oh and the last time I looked Sun has a Linux version yet I don’t see any Linux companies snapping at the bit to make their own Solaris versions when it goes open source. (Or asking sun to make their license more open)
There is a lesson to be learned in the fact that Sun is selling opterons, chips designed by AMD, a company that was once synonymous of low quality computers.
There is no “lesson”. AMD made a processor that didn’t suck and Sun decided to use it.
You mistake being a snob with being professional, I called users slack-jawed yokels not Sun. Sun builds quality much like a mercedez (sun) compared to mercury (dell). Do you think if Mercedes were to try and make cars for you and I that they would beat out Mercury? No of course not. But why would they try?
Whether the product is of professional grade or not doesn’t matter when it comes to the wallet.
It might not be a direct correlation, but they are most definately related. Sun’s sales come from their image as much as anything and were they to tarnish that by mass producing garbage for the average user, they would loose much of this. Remember, just because PC sales are driven by the lowest dollar doesn’t mean that the same is true here.
Sun might be hurting lately because people have decided to try and replace them with a PC that they threw together in 20 minutes which they install gentoo on with all of they’re custom settings. I suppose you can get away with this if you have the time to run around replacing hardware all day, but I know I don’t and I think many of the people that try to do this will find out that they really don’t either. Just because google pulls it off doesn’t mean it’s right
The only company making something out of BSD is Apple. This year alone a couple of BSD groups died. All that are left are Open BSD, Free BSD and Apple. Apple is the only one on that list that has a real product and that product is losing market share every day to Linux. Next time Bill Gates mentions that he is worried about BSD or Solaris the way that he is running wild about Linux you call me.
You forgot netbsd. Oh and yahoo does quite well too (in addition to apple).
Oh and by the way, as I mentioned before this is not a Linux thread why are you even here other than to start arguements?
Google is not the only company doing it:
http://www.techworld.com/opsys/news/index.cfm?NewsID=2736
“Plus Finanzservice, which provides services such as customer loyalty cards to clients including H&M and Ikea, migrated from Sun Microsystems’ Solaris operating system to Red Hat Enterprise Linux (RHEL), at a cost of 30 percent less than a Solaris system would have cost, Red Hat said on Thursday. Linux’s similarity to Unix, its lower cost and ability to run on Intel hardware makes the Unix market ripe for open-source conquest, according to many industry observers. Sun has been attempting to tackle the problem by partially embracing open source.”
Oh and by the way, as I mentioned before this is not a Linux thread why are you even here other than to start arguements?
I am here to rebut arguments that others have made, not to make new ones.
And and I would not say Apple’s PC unit is doing all that well.
And I will give you yahoo. But he main reason for that is because Yahoo started before Linux got popular. Remember BSD is much older then Linux!
I’ve never heard of Finanzservice before, so I won’t comment there.
I am here to rebut arguments that others have made, not to make new ones.
The 1st post you made did not refute anything, but rather made some claim that sun was dead like bsd. Now neither BSD nor Linux were mentioned in the thread’s title. So you’re arguement doesn’t really work. Though you have been doing ok since.
But he main reason for that is because Yahoo started before Linux got popular. Remember BSD is much older then Linux!
Yes, and Sun has also been around longer than Linux. Operating systems are like fine wines. They get better with age.
Dunno why people are willing to count GNU/Linux as being down and out… just because of Solaris 10 :/
Mind you this is from the same company who has lost so much ground to GNU/Linux, and the same company who made the OS that earned the moniker “Slowlaris.” GNU/Linux Systems has had the upper hand on Solaris for a few years in performance and equal in technology. (Personally I could never tell the difference in stablility between recent releases of Solaris, and GNU/Linux)
Some of you act as if these features will kill Linux off of high-end/middle/lower class servers… which is weird.
Its not like dtrace technology, if it proves effective, won’t be reimplemented in Linux; at Linus’s discreation. Or its probably not long before a filesystem is extended to 128 bits in Linux — though I personally don’t understand, which one of Sun’s well-known customers would need a 128 bit capable filesystem.
Linux has some interesting filesystems of itself, that rivial ZFS –GFS by RedHat; and Google FileSystem, which may or may not be released by Google Inc.
GNU/Linux Systems has had the upper hand on Solaris for a few years in performance and equal in technology.
Has had. This is changing in Solaris 10, and Sun continues to spend a large amount of time and money improving the performance of Solaris:
http://www.cbronline.com/article_news.asp?guid=2477BF5E-1C00-47F5-B…
Its not like dtrace technology, if it proves effective, won’t be reimplemented in Linux; at Linus’s discreation.
I’m assuming you meant discretion instead of dis-creation.
Related yet inferior technologies have proliferated for Linux. However, it is this exact discretion of Linus you describe that has lead to better debugging tools not being part of a mainstream Linux release. Bryan Cantrill, one of the authors of DTrace has blogged about this a bunch:
http://blogs.sun.com/roller/page/bmc/20040718#dtrace_vs_dprobes_ltt
Or its probably not long before a filesystem is extended to 128 bits in Linux — though I personally don’t understand, which one of Sun’s well-known customers would need a 128 bit capable filesystem.
I doubt these filesystems will take the administrative approach employed by ZFS. Do any of these integrate the disks, volume manager, and fs?
As far as the need for 128-bits goes, Jeff Bonwick, the author of the Slab Allocator, and now ZFS has a comment about this on his blog:
http://blogs.sun.com/roller/page/bonwick/20040925#128_bit_storage_a…
128-bits may not be ultimately necessary, but you’ll never need another larger filesystem.
I did mean discretion, thanks!
My post was primarily to show that Linux has been a contender and a capable OS for awhile, or atleast can compete with Solaris.
I personally don’t know what is faster, currently, however I anticipate benchmarks from independent sources should be around soon, if not already. I have not seen any comparing the latest GNU/Linux system’s vs Solaris 10.
I will take your word for it that Solaris 10 is faster than prior versions of Solaris, however, there is no reason to assume that Linux will stop innovation, and performance enhancements. As IBM, Redhat, and Novell all have kernel developers and also spend time and money on developing Linux.
I did add at the discreation of Linus for a reason, I’m aware of his omnipotent powers to reject patches, no matter how legitmate they now are.
Personally I feel dtrace is really neat,albeit, I feel it is slightly irrelevent if you can just see potential bottle necks if you are unable to have the source code to the operating system to fix them — assuming the operating system is the source to the bottle necks. Of course this is not a problem, if Sun fulfills its promises.
When I install my system on reiser4, I will never need another filesystem until reiser5 is released
*discretion
Personally I feel dtrace is really neat,albeit, I feel it is slightly irrelevent if you can just see potential bottle necks if you are unable to have the source code to the operating system to fix them — assuming the operating system is the source to the bottle necks.
Indeed, that is a substantial — and as it turns out, erroneous — assumption. Contrary to your assumption, we have found that the vast majority of bottlenecks — and all of the pathological ones — are not in the operating system, but rather in the towering stack of application software built on top of it; the gtik2_applet2 example described in our USENIX paper is representative of the kind of performance problem that we’ve found using DTrace. Our USENIX paper can be found here:
http://www.sun.com/bigadmin/content/dtrace/dtrace_usenix.pdf
”
I’m assuming you meant discretion instead of dis-creation.
Related yet inferior technologies have proliferated for Linux. However, it is this exact discretion of Linus you describe that has lead to better debugging tools not being part of a mainstream Linux release”
you dont get to manage Linux. Linus does. you cant give up control on Java nor tell Jonathan to stop lying about competition, so why try to find holes in what others do
If Linus doesnt want it, enterprise distributions can and do get these features patched themselves. thats the way opensource works.
do you even think openoffice only gets features you guys want. no way. it gets patches from everywhere. look how different it is from what you ship in openoffice.org
this is the way things work if you have unencumbered access to source code.
yes the same thing will happen with solaris if its open sourced.
I personally don’t know what is faster, currently, however I anticipate benchmarks from independent sources should be around soon, if not already. I have not seen any comparing the latest GNU/Linux system’s vs Solaris 10.
Last time I looked, it just depended. Sometimes Linux is faster, sometimes Solaris is faster. Often times, Solaris had more scalable algorithms that performed slower and needed to be tweaked to be fast for small systems. Performance is a constantly changing game. And everybody has an opinion about the viability of a given benchmark. It’s pretty contentious business.
I will take your word for it that Solaris 10 is faster than prior versions of Solaris, however, there is no reason to assume that Linux will stop innovation, and performance enhancements. As IBM, Redhat, and Novell all have kernel developers and also spend time and money on developing Linux.
That’s very charitable of you. I don’t often get such courtesy when posting. I agree that there’s no reason to assume that Linux will cease enhancing their performance, but in the same vein, neither will Solaris. Frankly, I think this is a good thing, as it drives continual improvement for both Solaris and Linux. Development groups at Sun have finally started taking the attitude that if something is slower in Solaris than it is in Linux, then it’s a bug a needs to be fixed.
To be clear, I’m not anti-Linux. I’ve run Linux since 1994, and still run it side-by-side with OpenBSD and Solaris.
Personally I feel dtrace is really neat,albeit, I feel it is slightly irrelevent if you can just see potential bottle necks if you are unable to have the source code to the operating system to fix them — assuming the operating system is the source to the bottle necks.
DTrace is more involved than that. Aside from the per-function entry & exit probed (FBT probes), there are also a wide variety of defined providers which have probes that measure more definable aspects of the system. These probes have an interface definition and will persist desipte changes to the functions in the kernel. So, it’s possible to write Dscripts without knowing all of the gory details about the implementation of the kernel.
That said, DTrace can also be used to debug user-level apps. You can use providers that let you instrument functions, instructions, etc in applications you run and observe their behavior too. Also, Adam Leventhal just wrote a new provider for DTrace which lets you instrument user-level locking primitives. So, it’s now possible to view the lock-contention in your user-level apps using DTrace. More on that here:
http://blogs.sun.com/roller/page/ahl/20040831#plockstat
So, there’s a bit more you can do than just look at kernel bottlenecks.
Of course this is not a problem, if Sun fulfills its promises.
Yes. They will. These things just take time, as Sun needs to work out a bunch of legal issues and eliminate encumbered code from the source base. I understand the skepticism, but Sun is going to do this. They’ve got people working on cleaning the code, and dealing with legal issues. It’s not a matter of if, so much as a matter of when.
I was refuting the FUD people put up saying Linux was this and that and Solaris is going to kill Linux etc.
“Yes, and Sun has also been around longer than Linux. Operating systems are like fine wines. They get better with age”
I never said that Solaris was not better then Linux, just like I said that BSD may be better then Linux in my posts before, but that does not mean it will sell better and it does not mean that Solaris or BSD will not die. Being that Linux in most lists has 4 % or more of the computer market (About the same as Apple) and BSD + Solaris together don’t even have enough market share to make a list. (If you don’t include Apple on the BSD list) I really don’t think you can compare the two (or 3)
Its based on MPL. Is it for OpenSolaris? Speculation, speculation.
The details, including a diff versus MPL, are here
http://www.sun.com/cddl/
“and the people at GNU seem to have little concern for making a good platform for proprietary software”
well, colour me surprised.
Intel make very little money selling chips to ‘good for nothing overclockers’, comparatively. They make a hell of a lot more selling Celerons to Dell. The *margins* on the ridiculous-overclocker market are great, but the volumes suck.