Guest post by After the recent removal of Solaris 12 from the Solaris road map inspired much speculation on the future of Solaris, Oracle has finally published a blog post detailing the cause of the removal, and the future of Solaris
Oracle Solaris is moving to a continuous delivery model using more frequent updates to deliver the latest features faster, while fully preserving customer and ISV qualification investment in the vast array of ISV applications available on Oracle Solaris 11 today. New features and functionality will be delivered in Oracle Solaris through dot releases instead of more disruptive major releases, consistent with trends seen throughout the industry.
In addition, support for current versions of Solaris 11 has been extended to beyond 2030.
The actual updated roadmap is light on details, though, but it does appear that Solaris at least isn’t dead just yet.
It’s to create new synergies with their already well developed continuous gouging model.
stop paying, updates stop. No keeping the system until EOL anymore
How apt, and that’s probably on the plus side with just an excuse for doing very little with the OS for the next few years before dumping it entirely.
Even if true not entirely sure rolling releases are a good thing for a server OS. Not corporate or admin of anything more than 6 or so PCs so maybe I’m wrong. It’s been known
Solaris will soon be as broken as Windows 10. And even better, now you have to pay for the privilege of having it broken! Woohoo!
Solaris is still a thing?
I have inherited a Solaris-rich server room with a new job. I’m a long-time Linux admin and will likely move away from Solaris over time. However, there are features of Solaris 10 and 11 that are way, way ahead of Linux. In particular, I’m just scratching the surface of what ZFS can do and I very much like what I see. Using it daily is pain compared to what I’m used to with Ubuntu or RHEL, but for where we’re using it, it’s worth the extra hassle still.
As for scrapping Solaris 12 and going with rolling releases, I like this new approach. I don’t care if they ever make a Solaris 12. Solaris 11 is a huge step forward from Solaris 10 with respect to patching, day-to-day use, and administration. I’m happier if they keep this version and just support it forever.
As with RHEL, I’m not aware of a supported (note I said supported) way to upgrade from one major version to the next. I’ll happily take stability and rolling releases.
Also, as for patching. Any professional users of Solaris in 2017 will have a support contract with Oracle. Even if you don’t for some reason, they have been pushing out patches after some delay for free. And to be quite honest, Solaris is very, very stable and is a very minor target these days, so delays in releasing patches shouldn’t really be a problem.
Rolling releases do not equal stability. If you really are a Linux admin, you would already know this. You did notice that RHEL/CentOS, SuSE, and Ubuntu are not on a rolling release schedule, especially where servers are concerned? There’s a damned good reason for that.
Woah. Relax.Not sure what set you off. Of course I realize that major Linux distros don’t do rolling releases and understand that rolling releases don’t guarantee stability. However, in the case of Solaris, much of their core is already more or less frozen in time. (Their JDS interface is based on Gnome 2 and they relatively recently removed CDE.) The core OS is quite stable and is unlikely to change much anyway. (for example: really, who needs to run a GUI on a Solaris server…)
As long as they put out new drivers for common server hardware, which also doesn’t change very much, they can keep going with 11.x for the foreseeable future.
Oracle clearly doesn’t want Solaris to compete with desktop or even server Linux. Ok, so that’s their call. I’m happy to have the current version supported forever and for Solaris to pretty much go in to maintenance mode, and I’m one of their paying customers.
I really, really don’t like Oracle as a company and this move obviously feels like they’re winding down Solaris development, but that doesn’t bug me much. I think they’re handling this uncharacteristically gracefully for them.
You could always investigate other systems for the ZFS support. For example FreeBSD, or SmartOS/IllumOS/Nexenta if you want to keep with an OpenSolaris-derived system. All have excellent ZFS support. You can also run it on Linux of course, but it’s a good bit less polished and usable than on the other systems, I use it locally but I’d go with one of the former for a production setup. I’ve been running ZFS on FreeBSD for over two years now on a NAS system at home and on vmware at work.
Absolutely! I’m already looking at SmartOS/IllumOS/Nexenta and am intrigued that Uubntu 16.04 included ZFS support.
I think that announcement is just a way of saying its in maintenance mode.
There is not a single new feature mentioned as upcoming. But there is a lot of talk about security updates and bug fixes.
Thats not a road map, its a cartoon map to flavor town on the back of a kids menu.
Oracle is laying off devs. I’ve been supporting SunOS or Solaris since 1993 as a professional, continuously (and still do). I worked as a C dev as well on Solaris and SunOS.
* ZFS, IPS, Zones & LDOMs, are cool, but too many embrace XML and displace plain text which is not a good idea and not the Unix way. You get dozens of great tools to work with text (sed, awk, etc..) and you want to cripple yourself with XML because “it’s structured” ? That’s a silly value system and proof of “doomed to re-implement poorly” playing out. Text is for humans, XML is for DNA sequencers.
* Dtrace == A lot of hype but still a fairly useful tool. Just calm down about it. Jesus, I’m tired of hearing about it. Other platforms have their own instrumentation & profiling solutions and you know what? Some of them rock, too, and are a bit less obtuse.
* Fishworks project (storage stuff based on ZFS) had/has some good ideas, but the implementation and launch was rocky as a mofo. I wouldn’t touch
that product ZFS or not. Lots of pissed customers on that front.
* ‘ipadm’ & ‘dladm’ == Broken buggy blaspheming garbage that would still be awful even if they managed to work properly. Give me my ####ing config files back. They should have sub-contracted Lennart Pottering to write that steaming pile of frustrating breakage. He’s got more experience than anyone I know at fixing unbroken things by way of god-awful ideas expressed in code. Avoid Solaris 11 if you like your network to pass traffic after trying basically any change on your interfaces.
* If I could combine Solaris 8 with ZFS, IPS, Dtrace, Zones/LDOMS, and remove SDS, it’d be about as good as you could get for Solaris, IMO.
* SMF == Absolutely horrible binary opaque blasphemy that works and feels like a hangover (SQLlite databases hold your configs). XML, etc.. It was an unbelievable step backwards. It literally turned me against Solaris and drove me fully into the arms of BSD (no regrets, either). It’s not that I don’t know SMF or understand it: it’s that I hate it eyes wide open. It’s not
the bugs or even the XML, it’s the *design* and the style. It’s mental nails on a chalkboard.
* I support Solaris 2.6 through 11. Still. Today. This very second I’m on a hot pager and someone could call me to solve their Solaris problems. I do it two or three times a week, at least. My opinion is that Solaris 11.Next is for suckers who drank two much corporate kool-aid. Get your yourself a 3rd party support contract for Solaris 10 or 8 from an American vendor who speaks clear English, patch to the hilt, version lock the system behind a firewall & NIDS and hold out until Armageddon with middle fingers facing outward. That is, if you can’t just move your application to FreeBSD and do yourself a giant favor.