I have been reading the CEO of Sun, Jonathan Schwartz’s blog, lately and it has some great material. I just love the fact that many CEO’s blog nowadays, as some of it is quite interesting, and bold, material. I suppose I would even go so far to say, that if I read a CEO’s blog and it wasn’t good, I wouldn’t invest in their company, as I would have my doubts about the leaders intellect and authenticity. On the other hand, if I read a CEO’s blog like Jonathan’s, I would be very interested in the company.
One of the recent posts that Jonathan made was in response to a flame by Linus. In his response to accusations, by Linus, that Sun was being disingenuous about truly open sourcing its ZFS file system, he mentioned that not only was Sun going to open source everything, but that they were going to do it under GPL3.
I think this is a truly brilliant strategic move by Sun, as it raises the bar for GNU/Linux and Linus who is very adamant about his dislike for GPL3. I do see there being competition between OpenSolaris and GNU/Linux in the coming years, and I do think it is a good thing, as competition is what drives innovation. I also see an interesting dilemma for GNU/Linux as OpenSolaris will be able to use GNU/Linux code, but GNU/Linux won’t be to use OpenSolaris code. This gives OpenSolaris a huge competitive advantage and might give them a temporary head start in the head to head competition of their operating systems. Ultimately, it seems like GNU/Linux might have a very large problem on it’s hands if it stays with GPL2.
I am not making a value judgement on GPL2 vs GPL3, but I will say from a strategic standpoint it appears to be checkmate for Sun on this round. Sun has some incredible virtualization and storage technology and it if enters the open source arena, “just right”, it could capture a massive amount of market share with this strategy.
Complaints by people in the GNU/Linux camp can easily be rebutted as Sun can claim the “moral” high ground as they support what the FSF recommends. I am quite interested in what happens over the next few years. There are quite a few possibilities, some of them admittely far fetched, but interesting to discuss regardless:
* OpenSolaris comes into the FOSS market goes GPL3 and takes significant market share as it has the best GNU/Linux technology and Sun Technology. GNU/Linux tries to fight back and switches to GPL3, but the damage is enough that they become the second preferred FOSS Operating System behind OpenSolaris.
* Nothing happens. Things pretty much stay the same and OpenSolaris doesn’t really make much of a positive or negative impact.
* GNU/Linux and Linus seeing the “end game”, quickly switch to GPL3 and cancel the advantage Sun might have before it happens. It then crushes OpenSolaris as it takes all of its good technology and the huge market share it already has and makes OpenSolaris marginalized.
* GNU/Linux becomes the preferred choice for companies that don’t want to become tangled in GPL3 and OpenSolaris becomes the preferred choice for a commodity operating system that runs in a data center. This leads to a significant loss in market share for GNU/Linux.
* Other things….what am I missing?
Comments?
Both growing in market share, mainly at the expense of proprietary Unix, but holding their own against each other.
IMHO distributions like RHEL and Debian have a major admin usability advantage over Solaris.
Other things….what am I missing?
One thing you fail to take in to consideration, is that when it comes to choosing your *nix flavor, It's distribution license, is often quite low on the priority list. e.g. there may be some sun only shops, that have heavy investments in Sun hardware/software stack, wouldn't it be easier for them to just migrate to opensolaris rather than Linux, GPL3 or no GPL3.
Same of Linux shops, not to mention the personal preferences of the System admins, in these places.
Whether Opensolaris beats Linux or the other way, I don't really thing it will be because of the choice of their License. If that was the case Free/Net/Open BSDs whould be clearly leading this race, as they are distributed under by far the most open licenses, and some may argue that they are also fat better choices than either Solaris or Linux.
I agree that RHEL and Debian based systems are known by many and that is a major advantage. People don't like to learn new things if they don't have to.
I am also curious about HP-UX and AIX. AIX in particular has some interesting virtualization technology too, but how long can proprietary Unix hang on? I really wonder if any *nix based OS will be proprietary in 10 years. It is almost like an OS is becoming a commodity and that being closed source on your OS is the kiss of death, at least in the server market.
bhasker/You raise a good point on the *BSD Flavors. What is going on with those? I guess I am unsure why their market share is so low still. If you read through the comments on Jonathan's post you will notice that at least one BSD developer was not happy as they felt Sun wasn't sharing their hardware design information fairly.
I am not sure though on the GPL3/GPL2 problem. If OpenSolaris goes GPL3, then it makes a HUGE ripple in the Linux community, as there are many people that feel so strongly, dare I say religiously, about free software, that it will become the ambassador of the cause. Also, as I mentioned that means that Linux would not be able to use things like ZFS, dtrace, virtualization from Sun as it would "infect" GPL2 code. I consider that to be very relevant.
If a piece of software is not up for the task, it's not up for the task whether or not it's GPL3 or not. And because even with gigabit ethernet, it takes a while to back up a terrabyte, not to mention petabytes. RAID and ext3 are not up to the task. You need ZFS or something like it. I already know people who are dropping x86 and Linux for FreeBSD on 64-bit, because that's where the future lies.
I also see an interesting dilemma for GNU/Linux as OpenSolaris will be able to use GNU/Linux code, but GNU/Linux won’t be to use OpenSolaris code.
uh, versions 2 & 3 of the GPL are mutually incompatible as section 6 of version 2 says you cannot apply further restrictions and version 3 is more restrictive than version 2.
http://www.gnu.org/licenses/gpl-faq.html#v2v3Compatibility
so OpenSolaris (GPLv3) is not going to be integrating Linux (GPLv2) code anytime soon.
the only workaround is what you see for GPL applications linked to openssl (which has a GPL-incompatible advertising requirement): an exception.
but exceptions are ugly because though OpenSolaris could then integrate GPLv2 code, it could not integrate GPLv3 code (unless that GPLv3 code also had a GPLv2 exception).
this is all mute if the GPLv2 code has the "or later" verbiage, but the Linux kernel doesn't, and the sheer number of contributors makes it difficult to relicense Linux (even to add "or later" clause). there are some theoretical ways of changing licenses, but as long as Linus is content to stay GPLv2, i doubt those will be tested.
I do not know about open solaris. Solaris itself should be a fine server os.
I have put solaris on desktop machines with gui and it was slower than xmas by a mile. Maybe the open solaris will cure that issue.
The reason sun could use linux code is because 30-odd% of it is under "gpl" (without a version, thus could be used in gplv3 code), and another 20% s licensed under "gplv2 or later"
@John: The Linux kernel is 100% pure GPL2. Applications and utilities in the distros are a different issue. There's nothing to stop distro builders using apps with whatever mix of licenses they like though, as long as they allow free distribution. The differences between licenses only apply when taking code from one app and putting it into another. For example it's possible to take FreeBSD licensed code and put it into a GPL app's source, but not possible to do it the other way around. However you can take a GPL licensed utility and put it into a BSD distribution without any problems because that doesn't require any mixing of code.
Just for the record I am using this quote from Linus as the basis for the GPL2 vs GPL3 debate:
"So to Sun, a GPLv3-only release would actually let them look good, and
still keep Linux from taking their interesting parts, and would allow them
to take at least parts of Linux without giving anything back (ahh, the
joys of license fragmentation).
"
Seems pretty straight forward to me...
hi ... another option is that both technologies and communities will thrive under whatever license they currently have or potentially change to. So, whether they compete or cooperate, no matter, both will grow well into the future based on their own standards of success. Is that a viable scenario? It's tough to judge, really. Linux has been open for a long time and has a large and mature global community, while OpenSolaris has only been open for a couple of years and is still closely tied to Sun. Those ties will weaken as the OpenSolaris community grows and diversifies globally, of course, but the community dynamics involved with OpenSolaris and Linux are so very different. The speculation is interesting, no question about it, but I have a gut feeling that many people are seeing this too much as a zero sum game. Perhaps it is and I'm the one missing it. We'll see. :)
Jim,
It is great to have somebody who actually works on OpenSolaris respond. I think you have a point about there potentially being a non-zero-sum outcome for GNU/Linux Communities and OpenSolaris. What I do find amazing is that it may not be a sustainable business model to sell a proprietary server operating system. I say this out of pragmatic analysis and not wishful thinking, as I still am not 100% sold on the concept of no proprietary software. I don't fully understand how the business model makes sense in every situation yet.
I really wonder what future AIX, HP-UX, and to some degree Windows Server have in an Open Source world that is embracing virtualization as Data Center 2.0. How can these Operating Systems possibly compete on price/performance value propositions?
hey ...
In the long term, I think those closed systems will find it challenging, especially when you consider the huge growth potential in emerging markets embracing FOSS (China, India, Brazil, Eastern Europe, etc). OpenSolaris is getting a warm reception in these new markets for two reasons: it's open and it's pretty darn good technology. But without the open part, it would be difficult to get back into universities and even to talk to many governments interested in FOSS. That's probably the number one reason why we opened Solaris (still opening, I mean, we're not done yet) in the first place.
Having said that, though, I do think that this will take a long time. There is a lot of proprietary stuff out there and more than enough time for HP and/or IBM to join the FOSS community with their respective Unix systems. Imagine AIX and/or HP-UX under GPL v2 so they could potentially mix with Linux. Or imagine HP-UX and/or AIX under an MPL-based license so that they could potentially mix with CDDL and OpenSolaris. Or imagine it all under GPL v3! Goodness, that would be a lot of code. :) Having been involved in the opening of Solaris, though, I do have a serious appreciation of the challenges involved in opening older code bases already under dozens and dozens and dozens of licenses. It wasn't easy for us, and IBM/HP would be in the same position.
Where do those moves put MS? I have my doubts that MS will open Windows, of course, but they are fierce competitors nonetheless. I live in Japan, and they are everywhere here. And Japan is far more conservative in embracing FOSS than most emerging markets are. The Japanese government does have programs to promote FOSS (and OpenSolaris is being implemented due to those programs), but again, these older markets are big and change occurs slowly.
I'm also not sold on the concept of "no proprietary software" at all, and I never viewed proprietary as always bad and open as always good. The global market is simply too big and complex for an either/or choice in all circumstances. There are some clear trends, though, and at least for many types of software in many critical markets, FOSS is the way to go to support an excellent corporate strategy /as well as/ an excellent engineering development methodology.
Moving to Japan from Silicon Vally has changed my views on all this substantially, by the way. :)
Simon, Linux license world is not so unicolor. All code, included into Linux tree, is licensed under GPLv2. However, this doesn't preclude original authors of particular parts of code to dual license code under other licenses. There are significant parts under GPLv2-and-later clause, there are parts GPL/BSD dual licensed. Also, Linus has the stance of "do unto others as they do unto you", so BSD code flows into Linux tree generally only with consent of original authors (and Linus respects, if they do not want their code to get relicensed under GPLv2).
So, in fact, there may be significant parts of Linux, which can be incorporated into GPLv3 code. Whether these are of any use for Solaris of course remains another question.
As long as my GNUish OS comes with all the standard GNU userland tools, I really don't care which GPL'd kernel it comes with.
I generally agree with most FSF recommendations and prefer the current FSF licenses, so if OpenSolaris went GPLv3 and there were a good distro for it (say, an "Ubuntu GNU/OpenSolaris"), I'd probably switch to that -- unless Linux were to also go GPLv3.
If both Linux and OpenSolaris were GPLv3, then I'd look into which one better upholds the spirit of the license, and also see which has the friendlier, more open and community-focused approach.
Linus Torvalds wants to license the linux kernel in GPL3 if Sun also does the same with Solaris:
"And yes, maybe ZFS is worthwhile enough that
I'm willing to go to the effort of trying to relicense the kernel[under GPL 3]." -- Linus Torvalds
Tabrez,
I did read that statement as well, but was not completely sure he meant GPL3. If so, it would seem to be a massive turnabout from someone famous for being stubborn and outspoken. In fact, it means perhaps he is admitting he was wrong? Does Linus do that? Interesting to watch...
The problem with your theory is that the kernel is the one program in which you have no choice but to hard code a lot of it. It is extremely difficult to port any significant part of the Linux to Solaris or vice versa. Linus pointed this out, in the posting to which you referred, regarding porting ZFS to Linux. Most of what Solaris needs anyhow is drivers, and I seriously doubt that Sun kernel developers feel they have much to learn from the Linux kernel.
The thing is, even if Linus strongly favored GPLv3, the kernel would not be relicensed for a very long time, if ever. This is because of the number of contributors who would have to agree to change the license. First they all have to be found, while for those who are not, their code has to be re-written.
But, it doesn't matter. Applications aren't linked with kernel code, they call system APIs. In that relationship, the license is irrelevant, which is why you can run proprietary programs on any GNU/Linux distro. And the main problem Open (or closed) Solaris faces is applications. In a recent interview, Sun's Ian Murdock (http://news.yahoo.com/s/infoworld/20070831/tc_infoworld/91468_1), pointed out that "The whole idea behind Indiana is to build more of a developer community around Solaris... How can we lower the barriers to programmers and run OpenSolaris as an ideal open-source operating system not originating from Sun?"
A lot of FOSS coders just have their personal machine for development and testing, not to mention doing their own sysadmining. And having written applications for commercial Solaris, I know it is different enough that it will not be easy for Sun to convince developers to install Solaris, much less port their apps. Also, I have to believe that Sun's ultimate goal is to run those apps on their own hardware, so where is that coming from? Having torn my hair out trying to get many FOSS apps to work on Solaris, even when provided by Sun, I can say that port is definitely not trivial.
But the biggest hurdle Sun faces in approaching FOSS developers is themselves. Their love/hate schizophrenia with the FOSS community has made many of us wary of getting too close. As Linus pointed out, what took them so long to free Java? And why won't they work with Eclipse? And what message was the payment to SCO suppose to send? They will have to do better than hire ex-distro developers to say, "We're not like that anymore."
Later . . . Jim
RenaissanceCore IDS, check it out at:
http://sourceforge.net/projects/renaissancecore
Before Solaris 10, I greatly preferred Linux. Solaris 10, I feel, has compelling reasons to prefer it to Linux. Such as ZFS and Dtrace. Linux has apt/yum for package/patch installs which is better IMO then Solaris. smpatch catches up the patches, but installing new packages still lags.
As a sysadmin, Solaris offers a stable API. I can take a Solaris 2.4 (1995?) binary and run it on Solaris 10. I can't do that on Linux.
If I want more bleeding edge, I can use OpenSolaris. Some of it has made it into Solaris 10u4 like iSCSI target too. Hopefully the next innovation will be putting the Companion CD that installs in /opt/sfw into an apt-get/yum like repository.
What are you missing? How about this one:
CDDL licensing model wins with OpenSolaris gaining momentum. GPL and GPL v3 become marginalized licensing models. Linux fades as yet another fashion fad.
15 years pass by. OpenSolaris displaces Windows in the desktop computing arena, having already secured dominance in the back-end, datacenter environment. OpenSolaris runs on many different types of hardware and powers appliances, effectively becoming ubiquitous...
UX-admin/That is an interesting theory. I am sure the FSF would be quite happy if that happened.
While the idea of OpenSolaris being GPLv3 sounds good, implementation could be quite difficult. If you've followed the recent and ongoing war with encapsulating BSD code inside of the GPL, then you know that even something as apparently as simple as the BSD license does not allow this. Leading to much frustration currently.
In fact, it is likely that Sun is already in violation with regards to several pieces of code just with their imitation open license they currently use.
I think the main point of Sun SAYING they are going to GPLv3, is purely political since they don't fully own (with regards to relicensing the code) the source in question and they are simply waving the GPLv3 banner since it would be difficult to move Linux out of GPLv2.
Your comments about "possible" future of/between GNU/Linux and Open-Solaris are quite interesting. As a user of both, as well as PC-BSD/FreeBSD I have often wondered about the great things to come with these Operating Systems.
The one possibility I feel is real, is that "if" Open-Solaris does significantly improve in areas where GNU/Linux has here-to-for excelled, and GNU/Linux does indeed fight to improve/stay ahead, then overall the scenario is a much better place for computer users in respect to *NIX versus MS Windows.
The one area that may prove very difficult for Open-Solaris is the "Embedded" field - for mobile technology, like cell phones and hand-helds, where GNU/Linux is hopping.
Very thought provoking
W.Anderson/You bring up a very interesting point. How would embedded GPLv3 OpenSolaris work? I know that Tivoization was a large firestorm that, from my uniformed seat, appears to have ignited GPLv2 vs GPLv3. Is there a sustainable business model that works when GPLv3 code is used in embedded hardware?
"I also see an interesting dilemma for GNU/Linux as OpenSolaris will be able to use GNU/Linux code, but GNU/Linux won’t be to use OpenSolaris code."
Please explain yourself as I don't understand.
Frank/I am not sure if you had a chance to see one of my posts a few comments up, but my understand of this situation comes directly from the mouth of Linus Torvaldes:
"So to Sun, a GPLv3-only release would actually let them look good, and
still keep Linux from taking their interesting parts, and would allow them
to take at least parts of Linux without giving anything back (ahh, the
joys of license fragmentation).
"
My interpretation of this quote is that GPLv3 can assimilate parts of GPLv2 code depending on how it is exactly licensed. I am not an expert on software licensing though, it is quite complicated to understand.
Linux switch to GPL3.
So Linux can include the OpenSolaris's technology! I hope!
If not, I switch to OpenSolaris!!!
"I also see an interesting dilemma for GNU/Linux as OpenSolaris will be able to use GNU/Linux code, but GNU/Linux won’t be [able] to use OpenSolaris code."
I don't think that's the case. The GPL 2 is inherently incompatible with the GPL 3. If OpenSolaris incorporates GPL-2-only code from the Linux kernel, it will be a violation of the licence of the Linux code.