We’ve just released SRU 20 for Oracle Solaris 11.4, the April 2020 CPU. It is available via ‘pkg update’ from the support repository or by downloading the SRU from My Oracle Support Doc ID 2433412.1.
The administrator of my organisation needs to supply me with a Support Identifier before I can do something as simple as read the documentation about this new version, so I have no idea what to tell you. I guess Solaris technically isn’t dead yet?
It’s nearly dead, isn’t it? Isn’t it basically just in maintenance mode now?
It’s in “buy this if you want fewer bugs in our Oracle database software” mode. Which was the whole deal with regards to the Sun acquisition.
So much money to make with Sun assets, yet no company besides Oracle bothered to actually buy Sun.
IBM was the only one that kind of considered doing that, but they withdraw their acquisition proposal.
This is something Oracle bashers keep forgetting, no one else wanted anything from Sun, not a single company, in fact Google most likely had a secret hope that Sun just closed doors and threw all assets into the dump bin around the corner.
moondevil,
I think there’s actually a lot of people who recognized Suns potential. They paved the way for so much of our tech. There are so many people who could have made better use of Sun’s assets than Oracle did, but none of us could beat Larry Ellison on the one thing that matters most in corporate acquisitions, which is money. That’s the sad truth. And it’s not just oracle, they’re just symptomatic of broad tends in the corporate world to get lazy and stop innovating. Apple’s wasted tons of opportunities to improve macos and when it comes to IOS they’ve sadly resorted to IBM’s model of increasing profits by locking things down rather than innovating. Microsoft was fortunate to become absurdly wealthy early on, but it’s spent most of it’s existence milking cash cows. Google and facebook are younger companies, but even there we see that once they reached a certain level of success, they cut back on innovation and focused on protecting their own incumbency and profits, wasting the opportunity to create something greater with their assets.
A lot of people who have the ideas lack the assets to change things, meanwhile a lot of people who have the assets lack the motivation to change things. It’s not impossible to find outliers, but with more and more wealth concentrating into fewer and fewer hands, it’s becoming rarer.
The reason nobody wanted to buy Sun is the fact that, under pony-tail dude’s supervision (Jonathan Schwartz), Sun went ahead with a variant of the Gnome Underpants Strategy:
1. Give the Solaris crown-jewels away to a competitor (Red-Hat) under open-source (with no way to translate the giveaway to increased revenue for Sun)
2. ???
3. Profit!
After the pony-tail dude was done, there was nothing in Solaris that customers couldn’t get from RHEL for cheaper and without the fear of lock-in to overpriced Sun hardware (remember, back then, people didn’t know if Solaris x86 was for real, or yet another diversion http://web.archive.org/web/20191002214345/http://www.sparcproductdirectory.com/view62.html ).
Unless as a company you were already knee-deep in Solaris lock-in, Solaris stopped making sense. Sun became a provider of backwards-compatibility and nothing else, much like Unisys. Only without those lucrative mainframe contracts and obscure mainframe architectures (migration from Solaris was and is easy, especially after all that open-sourcing of its bits).
Basically, the pony-tail dude open-sourced the crown-jewels without any long-term plan whatsoever. Good for open-source, bad for Sun as a company.
Feel free to draw any parallels in your mind about that other suspected trojan horse of a CEO, Stephen Elop of Nokia. But at least Stephen Elop had the scuttling of the mothership planned ahead and was handsomely rewarded for the act, Jonathan Schwartz didn’t got a penny from Red-Hat or Oracle.
Basically, the story of Sun under Jonathan Schwartz should be a cautionary tale for every open-source-friendly CEO out there, that if you want your company to contribute to open-source, make sure it’s either done in a way that’s profitable for the company (Google’s Android), or that you don’t have any stockholders to please in the first place (Canonical). Don’t just take a public company (that’s meant to please stockholders) and try to make it into an open-source powerhouse in a way that doesn’t benefit the stockholders, and without any long-term plan.
If you do, you will be voted out of the CEO throne by the stockholders and the company will become the lunch of much bigger fish.
If I’m not mistaken, Sun was already hemorrhaging before they started dabbling with open source. It was more a kind of desperate actions what they did but, true, deep well-thought-out it was not.
@kurkosdr, What on earth are you talking about? What parts of the Solaris crown jewels can be found in RHEL? DTrace can only be found on macOS and Oracle Enterprise Linux. ZFS is nowhere near Enterprise Linux and almost made it to macOS. Zones, while a great concept and way ahead of it’s time were kept in Solaris. It took another 10 years before we finally had docker. SVCS stayed in Solaris. No Solaris IP ever entered RHEL.
The only thing that Red Hat wanted was the iPlanet Suite (mainly the Directory Server and the Certificate Authority), codeveloped by Netscape and Sun Microsystems (later named Java Directory Server). And it bought those from AOL (the owners of Netscape at the time of the purchase). They are known today as 389 Directory Server (foundation for FreeIPA) and DogTag Project.
SVCS was refused in Fedora after a long debate due to licensing (apparently CDDL was not acceptable, but it’s twin MPL was) and a SVCS/LaunchD replacement was created that we call SystemD after a short attempt to use Upstart in RHEL6. DTrace wasn’t even considered. ZFS was refused for the same licensing reasons, even though there was a read-only GPLv2 implementation sponsored by Sun Microsystems in GNU GRUB, so the IP issues were quite clear. Comstar (the storage framework) wasn’t even considered. So, I ask you again: what part of the open source Solaris did you see in RHEL? GNOME 2.0? That was a joint effort by Sun, Ximian, Eazel and Red Hat. For Sun it meant getting rid of the ageing CDE that they maintained exclusively and using something that shares maintanance between multiple vendors. For Red Hat it meant a modern desktop that had no potentially risky IP (like QT does). For Eazel it meant bankruptcy. For Ximian, the company of the original GNOME creators, it meant they were purchased by Novell and ready for a refocus to Mono/.NET technologies.
The mistake was made in the Solaris 9 era when they wanted to kill the free (as in beer) Solaris x86. Solaris x86 was their entry into new business. Any new admin could download Solaris from the Sun website and could install it, thus learning it and evangelising it. They quickly reverted that mistake and decided to invest in Solaris 10, but it was too late.
The ponytail strategy was quite smart, but came in too late. The secret to the mid 2000s UNIX was to be legacy-free and free as in beer with expensive support contracts for commercial usage. Need I remind you that while Solaris is barely alive today, it is much more alive that AIX, HP-UX, Tru64, UnixWare and many others?
The free as in beer worked and gave Solaris a few more years to fight, but it was difficult to adopt by the emerging Administrators. UNIX Administrators were a few thousands world-wide. Linux Administrators were almost by the millions in 2005. The growing market needed administrators and they used what they could find: Linux Admins. While much more limited in knowhow than their UNIX counterparts (back then), they were at least available. In order to get Solaris out there, they needed to make Solaris much easier to adopt and that meant getting rid of the cruft.
OpenSolaris was their break from the legacy constraints. Solaris administration was incredibly difficult for non-Solaris folks and OpenSolaris (what is now known as Solaris 11) was the break away from compatibility with the legacy userland. Solaris 10 had ucb compatibility libraries from SunOS 4, the predecesor to Solaris from the late ’80s. It had NEWS (pre X11) compatiblity, it used Display Postscript. Almost any Solaris application compiled for any Solaris version (32 or 64bit) could run on Solaris 10 64bit and that was difficult for the Sun Microsystems teams to provide. OpenSolaris came with a new FHS, with new utilities, with a complete modernization of the userland next to the power of the Solaris technologies (that are still quite impressive today).
During the OpenSolaris development cycle they actually did something even smarter for Solaris 11, they produced both the SXCD/SXCE builds (Nevada) which were the new technologies on the old userland and the OpenSolaris distribution that was legacy-free. The Nevada builds were incredibly similar to Solaris 10, but included every new technology; they used the old package system, the old Filesystem Hierarchy Standards, the old Java based installer, the old library compatibility.
There was a time I was very excited to try out Solaris on my machine. It worked okay with the CDE, but getting something modern would be nice. There was actually community enthusiasm around the project.
But the CDDL was crafted to prevent Linux from scraping parts from Solaris. What they did not realize is that the opposite was more likely. Linux had much better support for hardware, and also had modern graphics components. They could not even benefit properly from FreeBSD either.
And the final nail was Oracle purchase, and shuttering the open source project. Even IBM is now using a hybrid AIX/Linux strategy on their systems. They could have made better choices.
ACtually, for Oracle, it was NOT a bad choice. The target was JAVA. WIth ip rights to Java, they basically owned the internet world. Solaris, ZFS, etc. were only very minor players in the money scheme. As for Solaris, it’s only alive as long as Financial clients such as banks, still purchase support. They can afford it and provide teams of in house DEVs to support it internally. As for graphics, modern hardware, etc. That is only gifts for the non paying user. For a large soulless organization such as Oracle, Joe User doesn’t matter, and loss of them is inconsequential. I don’t condone it, but I understand it. If anything, we need large paying organizations such as J.P. Morgan, Amazon, Bridgewater Capital, Facebook, TD, Google, and Blackrock Capital to hedge their support contracts in a way that will leverage more community friendly support. But the odds for that to happen are negligible at best.
Oh, how I miss Sun. If I remember correctly, they were considering licensing Solaris under the GPL3 as well as CDDL at one point. It would have produced a rich ecosystem – and here we are today with a barely maintained OS.
Former Solaris QA engineer here, I worked for Sun / Oracle until pretty much fairly recently when they pushed me out of the door (I work on Solaris for about a decade). I was a True Believer, Solaris was such a beauty internally at so many levels, it was highly innovative when the development was still active, that is until when Solaris 12 was still under development. Working with these teams was a delight, many very passionate and competent people, fairly elitist folks but not minding where the good ideas came from! Even new recruits.
Now I can tell you for sure Solaris is zombie-level dead. Concretely almost all the engineers working on it have been fired. An s11u4 SRU is pretty much a bunch of security fixes + open source user space updates (python, etc). Basically it’s just maintenance, nothing more.
IMHO the main reasons that doomed Solaris in the OSes landscape are:
1) it failed to become opensource soon enough
2) customers paying for it were mainly seeking reliability
For 1), Linux killed it, and as far as I can see, Linux is slowly but surely killing all the other forms of Unixes, Linux is were lies true innovation today and where the attraction is for new users coming out of schools etc. Many youngster have barely heard of the BSDs and there is less and less reason to go there unfortunately.
For 2) many of the customers did not upgrade their versions, why upgrading the OS when the old thing still works super reliably? That was definitely a problem for Sun / Oracle that had difficulties to justify upgrades. One of our customers call us for a problem on a system, the engineer routinely checked the uptime, the thing never had a shutdown in … 12 years of continuous uptime! (I don’t know what services it was running though to be honest).
Yes I greatly miss Sun 🙁 I always feel a tad sad and nostalgic when I think about it.
I agree. Solaris and OpenSolaris were wonderful operating systems. I used Solaris 10 instead of Linux at work. Guys thought I was crazy, Linux was “faster” in their mind. Then they tried to run the same workload on their workstation as I was running on mine and they couldn’t open a shell because Linux was overwhelmed, and my box just kept on chugging along without noticeable difference. Sure it was slower when you weren’t doing anything on Linux, but once you put a load on both Solaris just kept going. It was great.
I made the journey to OpenIndiana after Sun shuttered OpenSolaris. Then went to Mac. Now I’m probably going back to Linux or Windows for the desktop as I’m not happy where Mac has gone.
The best operating systems sadly just don’t seem to stick around. I miss OS/2 and Solaris. Yes, both are still available in some form if you’re willing to pay for it, but they’ve lost so much community mind share that there’s not enough going for them these days.
Oh well.
I used Solaris as a primary OS for quite a while. I started out with SXCE, later switching to Nexenta, and then briefly used Tribblix, before switching to FreeBSD (I still have VMs of my Nexenta and Tribblix installs around). The biggest reason I was using it was because of ZFS (I had it on ZFS root before that was officially supported). Generally I found it to be fairly reliable IIRC (it’s been a little while).
And yes, the state of OSes is pretty sad right now. I think the only way for an alternative OS to survive is to have strong Linux compatibility (which is what I plan to do with the OS I’m currently writing; it will be compatible with Linux applications and even most Linux drivers despite being a lot more like QNX than Linux in terms of general architecture). None of the current alternative Unices seem to get the importance of Linux compatibility for some reason.
I completely agree with you on the importance of Linux compatibility, all the cool opensource projects nowadays are Linux ones and most of them do not care much about the compatibility with Unixes as it was the case before.
andreww591,
Simply put, it’s popular because it’s popular.
When the market is growing (like when microsoft and linus started out), you can build a new OS to fill a void. When the market is saturated and even shrinking, the only way for your OS to grow is for you to displace another incumbent, which is poses significantly higher growth challenges.
It kind of sucks, but the era of huge demand for new innovative operating systems died with the tech collapse two decades ago. Of course there were momentary voids in the mobile market that might be an exception, but only temporarily, that market is a solid duopoly now. The vast majority of consumer attention & cash goes to the very top and unless you’re working on one of the top major platforms, you are irrelevant. 🙁
There was someone on osnews years ago I think you would have really liked named Neolander. We would enjoy a lot of technical discussions and he even posted an OS development series. I wish we had more content like that. We talked about OS dev because it was interesting and fun, but even ten years ago we knew it was too late to change the market. I started osdeving in the early 2000s but moved onto linux when it became apparent the market was dead to newcomers. Maybe I could have done something to be compatible with linux, but then what’s the point? My passion lies in making things better, not making copies.
http://www.osnews.com/story/24270/hobby-os-deving-1-are-you-ready/
As you can see, I’m not optimistic about the prospects for alternatives. I still really enjoy talking about the tech though, so if ever you want to talk about it I’d like to do that!
Oddly enough, now might be the time.
Redox is interesting in how it’s implemented in Rust, and the explosion of cores (64 real cores on a CPU package!) makes it time to rethink how we view kernels and system resources.
In a couple of years, CXL and Gen-Z interconnects are going to go mainstream which means things are going to look like Plan9 at the hardware level.
We’re at one of those peaks where everything changes. It’s like a lizard molting. We’ll look back and think, “We were so primitive back then.” This, of course, is happening on the server side, where people will shift paradigms if it means results. The consumer side is trash, and it always will be. Google shifting Android’s base away from Linux will the be biggest change we’ll see there.
Flatland_Spider,
I think you’ll understand the reason for my skepticism. Looking back, we’ve seen many changes in new markets with the most opportunity for new players. But in old markets it always ends up consolidating down to just a few viable players at the top.
Nevertheless allow me to raise a glass to the alternative platforms of the future, cheers 🙂
The Hurd kernel could use some work, and if you could get that into some sort of production ready state people would love you.
Most of the current OSes in the *nix universe are contemporaries of Linux or predate Linux by quite sometime. Linux is the new *nix on the block, so Linux compatibility wasn’t a things when the others were built.
With that being said, FreeBSD does have a Linux compatibility layer for userland applications. If you’re hellbent on writing a Linux compat layer, you could start by helping them with that.
There is also the problem of the GPL. Many OSes aren’t under the GPL, but by linking to GPL drivers, their kernels would come under the purview of the GPL. Of course, this doesn’t stop anyone from stealing the code.
Also, many of the OSes port the stuff they can from each other. FreeBSD has Intel video drivers because of the way Intel licensed their code. Applications get ported all the time between different OSes, but the application projects also can reject patches for other OSes requiring the porter to carry out of tree patches to make the application work on their OS. This happens a lot to the BSDs. The upstream project will reject patches for their OS, and the porter has to maintain patches for the application resulting in lagging versions.
Overall, it’s not as simple as “they don’t understand.” It’s about politics, project vision, ego, and resources. It might be good for you to step out of the Linux world for a while and see what problems other OSes run into.
There’s way too much wrong with GNU/Hurd in my opinion to bother contributing to it (the biggest problems being Mach and the translator model of filesystem servers, both of which are fundamental to its architecture). QNX’s architecture is far superior to the Hurd in every way. If QNX were open source, I might have contributed to it, but since it isn’t, I’m writing my own OS with a similar design that improves on that of QNX in many ways.
I know that most other Unices either predate Linux or started around the same time, but if they want to stay relevant, they have to adapt to the new reality of Linux world domination (I’m not saying that it’s a good thing, but it’s just the way things are).
I’m well aware of FreeBSD’s Linux compatibility layer, but there’s been a lot of stuff I haven’t been able to run under it. It seems as if it’s not much of a priority to them.
Reading my reply this morning I realize a few key words were left off my final thought. It should have said:
The best operating systems sadly just don’t seem to stick around. I miss OS/2 and Solaris. Yes, both are still available in some form if you’re willing to pay for it, but they’ve lost so much community mind share that there’s not enough going for them these days to justify the price..
Thanks for this wonderful insight!
Solaris had some great tech and engineering! It was amazing, and the SmartOS derivative is still being really innovative.
However, it wasn’t as user friendly as Linux or the BSDs, and that was a big part of the problem. People will talk about how unfriendly Linux/BSD is have never sat down in front of the Solaris box. The Linuxization of Solaris effort was too late.
There are many reasons to go with the BSDs, but they aren’t the kind of reasons people run into unless they’ve been working in the industry for a while.
Slowly, but surely, Linux distros are running into problems which will need to be addressed by significant refactoring, and the solutions have the possibility of really annoying the hobbyist user base.
Yep, that sounds like Oracle. They cling to a very old model of “enterprise support” that even Microsoft abandoned years ago.
I worked for a company which was acquired, and locking up the docs is one of the first thing the new overlords did. Needless to say, I did not last very long after the acquisition.
I can understand locking up proprietary, internal documents relating to implementation specifics. That’s no different than having the source closed. I cannot understand the desire to lock product documentation such as manuals, security advisories, release notes, etc away. It has never made any sense to me.
It was even dumber when you consider the docs were for an open source based product.
Locked up docs/bad docs/no docs is one of the top three reasons I eliminate software from consideration. It’s right after being Java or Ruby based.