Early December of last year, I posted the rumour that Oracle was going to end Solaris development. While the company denied these rumours at the time, there still seems to be something going on.
Rumors have been circulating since late last year that Oracle was planning to kill development of the Solaris operating system, with major layoffs coming to the operating system’s development team. Others speculated that future versions of the Unix platform Oracle acquired with Sun Microsystems would be designed for the cloud and built for the Intel platform only and that the SPARC processor line would meet its demise. The good news, based on a recently released Oracle roadmap for the SPARC platform, is that both Solaris and SPARC appear to have a future.
The bad news is that the next major version of Solaris – Solaris 12 – has apparently been cancelled, as it has disappeared from the roadmap. Instead, it’s been replaced with “Solaris 11.next” – and that version is apparently the only update planned for the operating system through 2021.
Read into that what you will. Sounds like maintenance mode to me.
Looking at the roadmap, I wonder if this means Solaris is going into maintenance mode or maybe Oracle wants otm ake it more of a rolling release. Microsoft announced a while back there would be no Windows 11, just constant updates of Windows 10. This move, with the name “Next” feels similar.
You are right. It is all about naming not technologies. Solaris 12 just become Solaris 11.4. Do You remember when Solaris 2.7 overnight become Solaris 7. Now it is similar process but somehow reverse.
Take a look here – https://forums.theregister.co.uk/forum/containing/3078659
Yeah, because when a software hits “50.1.0”, with almost weekly updates, it doesn’t sounds like it’s a mastered project as a whole.
I rather like it – you need some way to keep track of versions, and Mozilla doesn’t promote Firefox based on it’s version number any more. It just becomes mostly internal book keeping, which is fine.
I think most people are worried about the layoffs involved instead of the name change.
Let’s be honest, Solaris was a goner even before Oracle took charge. Nobody really believed the whole OpenSolaris thing, because TheRealSolaris that the system they bought actually runs is a closed and deeply proprietary system, and the wounds from UNIX lock-in are still fresh (if not still getting created). So, CentOS and RHEL just replaced Solaris. Jonathan Schwartz’s idea to keep the delivered product closed while at the same time open-sourcing some of Solaris’s crown jewels for RedHat’s pleasure was a disaster for Solaris. Oracle just accelerated the decline with their whole “Complete systems, engineered to lock you in together” initiative.
Now, on SPARC… Assuming you still need to buy iron instead of renting storage and computing from Amazon, SPARC got its lunch eaten by Intel and AMD because the chips were better value per buck. Plain and simple. IBM’s POWER kicked it out of the HPC niche. The smart folks who build Top 500 supercomputers don’t choose SPARC anymore and that tells a lot. SPARC got its butt kicked by ARM and MIPS64 in embedded because SPARC mandates the needless replication of registers for the mandatory and mostly useless “register windows” feature and just couldn’t compete.
So, I can blame Oracle for a lot of things they did to Sun, but Solaris and SPARC were goners anyway.
Edited 2017-01-19 12:26 UTC
Regarding SPARC, Niagara was a big hit and Sun sold thousands of them, SPARC was doing pretty good back then… BTW I give you the benefit of the doubt with SPARC because (IMHO) it was never a killer product and “the Cloud” is definitively Intel/commodity-based… but I totally disagree with you regarding Solaris.
Sun cared a lot about Solaris, even in 2010 Solaris 10 was a wonderful super innovative product getting better and better with every release, the day Oracle bought Sun, Solaris was the best enterprise OS in terms of technology and innovations and Open Source communities around it were thriving (thanks to the adoption of OSS/libre lics). You could like Solaris or not, but in 2010 It wasn’t a dead OS at all, It had a rock solid fanbase and a huge community for an Enterprise product (how many AIX or HP-UX fans did you know? Solaris was super important even for Linux fans).
7 years later, Solaris is in zombie mode, there’s no community at all and RHEL is the “Unix” king!
These are facts not opinions. Only Oracle can be blamed about this bleak Solaris present because it’s the result of their decisions.
Edited 2017-01-22 18:56 UTC
maybe oracle prepare a clould version of solaris…
hahahahaha! Best comment ever, and i wish i could upvote more.
https://www.youtube.com/watch?v=0FacYAI6DY0
“it is the most nonsensical term ever designed in computing”
Edited 2017-01-19 06:51 UTC
Half a decade from this:
http://arstechnica.com/business/2010/12/oracle-launches-sparc-super…
Solaris is still the only OS that has ZFS encryption (fs-level). Even automount using an external key (over https) is possible. I like it a lot.
Maybe HAMMER2 (DragonflyBSD) will be the future for my use-case.
https://github.com/zfsonlinux/zfs/commit/0b04990a5de594659d2cf204589…
evert need to reword you statement its slightly out of date.
Solaris is the only OS that has stable ZFS encryption at this stage. Zfs on Linux has not fully audited prototype ZFS encryption. So maybe in 2017 ZFS encryption will be production ready on Linux.
Of course for ZFS encryption to be production ready on Linux the full partition keys of ZFS being storable in LUKS is kind of a requirement.
Automounting with a remote key file for encrypted drives that is something that can be done under Linux or Freebsd but is horible. If this is a feature you like it something to at least put in a feature bug requesting on be it either Linux or BSD. Do remember hadoop if someone says we are not doing key files remote due to some security reason. Yes hadoop supports http and https as well as being a fuse mountable file system that you can stuff a keyfile into. Both freebsd and linux has definable encryption keyfile path for auto-mounting drives so key file over http is doable just its super big hack.
https://gitlab.com/cryptsetup/cryptsetup/
https://github.com/shpedoikal/tpm-luks
Under Linux it would be a feature request for cryptsetup or independent project to make it neat and tidy. BSD I cannot remember that the particular parts are called.
There is always a mix up between what is a unique OS feature and what is something that has just been made simple. Linux and BSD both can be made do keyfiles for auto-mount drives over https neither make the process user-friendly. Solaris tools are more user-friendly to do the https keyfile. So if this is a critical feature to your operations placing some feature request to either the BSD tools or the Linux tools would be a good idea.
Remember when something can be done horible this does limit arguments against implementing the feature in a better way.
FreeBSD uses GELI for its disk block encryption.
There is also a OpenZFS 2016 Summit video talking about the ZFS encryption implementation. Video is here:
https://www.youtube.com/watch?v=frnLiXclAMo
I think some of the Illumos vendors Delphix / Datto are already rolling it in for future releases.
I haven’t seen anything regarding FreeBSD picking it up, but I haven’t been looking too hard either. However, I hope they do.
Thanks a lot for the info!
Ignore how FreeBSD has ZFS and geli natively, and sure.
Edited 2017-01-21 08:25 UTC
Ignore how I talked about filesystem-level encryption. Geli is another layer.