snw's post might be reflecting the fact that OpenIndiana isn't seeing a lot of updates (except perhaps for those that come directly from Illumos.) For example, when the heartbleed bug came to light, it turned out that OpenIndiana wasn't vulnerable, because the version of ssh they had was so old it had never had the vulnerability introduced into it in the first place.
As yet I don't have any hands-on experience with OmniOS, but I'd tend to agree that it seems the best direction to go looking in, for a conventional OpenSolaris-derived OS.
this is indeed the biggest reason. They have a "hipster"-branch [1] that is getting some updates but I don't follow that closely. Another reason is that they try to target the "developer desktop" which is not that interesting to me personally and also probably a lot more work (driver wise) than they might have the manpower for...
OmniOS on the other hand is pretty solid and with pkgsrc (default package system on SmartOS and NetBSD) you get nearly 13k packages of open-source software.
My List is also missing a few other commercial players that use and contribute to Illumos:
- Delphix, uses ZFS snapshots to do fancy things with big databases
- Pluribus has "Netvisor OS" that is some SDN Router/Networking appliance [3]
And then there is the long tail of small hobby distributions:
- Tribblix, "Retro with modern features" [4]
- DilOS, which has a focus on Xen and Sparc [5]
and probably more that I forgot and don't know.
It is a very nice community and lots of interesting technology gets developed.
Our company migrated nearly all servers from Linux to SmartOS in the last 2 years and we could not be happier.
Thanks, that's helpful! I'm mainly experienced with Debian, but I've been exploring whether Debian-for-everything is still the best policy. In terms of actually trying anything else, I've only been doing some test setups of FreeBSD so far, but "something Illumos" has been on my radar as well and I've been trying to understand the landscape there, along with playing a little bit (not very seriously, I'll admit) on a SmartOS instance through Joyent's cloud.
Technically it seems quite impressive, and Illumos being more or less the ZFS upstream is appealing. The distribution situation has been more confusing, though. OpenIndiana seemed to have a community, which is why I was first looking into that option, but it does indeed not seem to be very active. That's one attraction of FreeBSD & Debian, that they're community supported, and I can with fairly high confidence expect they'll be around in 5 or 10 years supporting the distribution. It's less clear to me how to wade into Illumos in a way that mitigates risk of the upstream going away. I think Illumos itself fits that description, with a community that will be around, but does any individual distribution? There are clearly resources behind OmniOS and SmartOS, which gives some confidence, but they are also quite heavily concentrated resources (one company drives each, and it's not impossible that they could change focus and deemphasize development of the public distribution).
Thats a very valid concern. For SmartOS it would be a major blow if Joyent were to pull all ressources. They have some of the brightest engineers I know working on it and I have learned a lot from them by just using the system and hanging out on the irc channel in freenode.
Without them the speed of development would definitly take a hit.
On the other hand they have been - and still are - very good at building a community around it. There are multiple people and organisations building their own SmartOS images/derivates ([1], [2]). Pull-requests on github come in from a diverse enough group that I think it would survive even without joyent.
For OmniOS I don't have enough insight to make that judgement. They have a very clear and detailed release plan and recently hired some good people just to work on OmniOS. These things definitly add some confidence but of course is not a guarantee forever. But if that would be needed one can always buy support from them.
