“Oracle today announced Oracle Solaris 11.1, delivering over 300 new performance and feature enhancements to the Oracle Solaris 11 product family.” This stuff goes way over my head.
“Oracle today announced Oracle Solaris 11.1, delivering over 300 new performance and feature enhancements to the Oracle Solaris 11 product family.” This stuff goes way over my head.
Honestly Thom, I’ve been reading your articles here since you’ve started so I think I have a fair sense of your depth of knowledge of OS internals — there’s very little in the way of Oracle’s press release that goes over your head; most of it is just Oracle tradmarked jargon and product branding. Spend 45 minutes with Wikipedia and you’ll understand more in that press release then they guy that wrote it.
=O)
Edited 2012-10-04 22:02 UTC
I have an adequate knowledge of Solaris 9 and 10, and struggled a while with Solaris 11 before abandoning it completely.
Call me sceptic, but is there a link to those 300 enhancements ? Not blah-blah-17%-cloud-performance-gain, but the detailed technical list.
That number just refers to the number of items closed in their issue tracker. This is the industry standard absolutely useless number.
I’ve been working with Solaris 11 for SPARC last months and I really like it. Much more user friendly than 10 for people with Linux background… GNU toolchain works by default, bash is the default shell and the new command line utilities are nice. Cool.
The only drawback is SPARC support, you need a sun4v to use it (or x86-64)… ’cause Oracle dropped support for every previous SPARC system even the IV+ ones! It’s a shame… but hey, We know Oracle, they’re a bunch of blood suckers and they’re proud of it.
BTW I think Solaris is the best and more modern enterprise Unix out there by a large margin, it’s stable, it’s easy to use and It adds cool and useful features every release.
And the funny thing is, If Oracle integrates ZFS into Oracle Enterprise Linux they will have the best enterprise Unix and the best enterprise Linux too! Oracle the best *nix vendor in the world!! OMG!!! xD
I still remember the days with Solaris 2.4, where my first configuration step was to install all the available GNU tools from Sunfreeware.
Otherwise I would be stuck in the 70’s, System V style.
Still it is good to know about the new version. Many of our customers only care about commercial UNIX systems for the critical parts their infrastructure.
There is plenty of money to be made in projects with Tru64, HP-UX, Solaris and Aix.
I’m just glad that Oracle are pushing Solaris. Then they took over Sun, I had a horrible feeling that Solaris would be left to rot in favour of Oracle Linux.
Thankfully I was wrong
I guess that was just OpenSolaris.
Edited 2012-10-06 22:50 UTC
Well yeah, but I wasn’t as worried about OpenSolaris, partly because it was pretty naff* and partly because you can’t kill open source projects of that size.
* that said, the various OpenSolaris forks seem to be showing great potential lately. I quite like the look of SmartOS, for example.
solaris is so good it is an absolute tragedy that it also sucks.
thom, all you need to know is that the meager changes in solaris 11.1 are closer to being stripped from its bones by the FOSS community to give illumos a bit of extra juice like in “the mummy” when the guy who looks kind of like billy zane but isnt eats the people who woke him up. arnold vosloo gonna fk oracle up. that’s what you need to know.
Most commercial UNIX systems suck, if you compare them with Linux and BSD. At least in user friendliness.
When installed out of the box, most still seem to live in the 70’s, early 80’s.
Still, they are quite robust, and they even offer stable APIs!
Stable or stale ? 😉
Just kidding, couldn’t resist.
That news means more work for me. I do Solaris deployments on a daily basis. Its basically my bread and butter. I think Solaris 11 is just the SRU 10.5 now to 11 and integrated but i havent seen any changes in the ISO files so it might be not yet released.
As for the OS itself.. i hated it at first but got to like it. The good stuff (Network bonding, ZFS, COMSTAR) is the most easiest thing to do. A few things like ipadm and such were a bit hard to get used to, but all in all its a very good unix. 11 is quite a improvement over 10 and since i only work with the newest Oracle / SUN Hardware i couldnt care less about the dropped support and seriously: About time you drop all the bloated legacy hacks and start somewhere from scratch. Linux did the same with my beloved Hardware (Like the fritz card b1 and its hard to get it working in any recent kernel).
Oh BTW, if you want a nice alternative to Fishworks (Basically a GUI to ZFS NAS based on Solaris) you might want to try Napp-it. It works also with SPARC.
Let’s stop talking about “stability”, “scalability” and “performance” in vague terms. That’s what the salespeople do.
In 2012, what do we mean exactly by “performance”, “stability” and “scalability”. In what way is it better or worse than, say, Linux or Windows?
Performance – let’s start grounding it – do we mean disk performance? Do we mean network throughput? Latency? Do we mean performance of the core maths libraries?
Scalability – again – do we mean ability to scale with larger memory? Concurrent threads or apps hitting the network io? Memory manager and task scheduler coping with growing number of processes? Threads?
Stability – this is a funny one. I’ve never seen a good definition when people enthuse about their favourite OS’ stability. Do they mean uptime left on its own? Do they mean not losing data when hit by a web-request storm? Do they mean ability to keep lots of processes happy and not kill them when memory is starved?
Do they mean that the level of bugs is low – so things don’t crash or restart or just not work? Maybe we should measure “correctness” of an OS instead here?
The last time I say any attempt to quantify and compare such things was about 10 years ago http://bulk.fefe.de/ — we don’t see reviews of OS releases like that anymore.
I can’t believe 80% of Linux reviews are reviews of how pretty the installer is, the desktop theme and the default package selection.
+1
That drives me nuts as well! Those are the least significant parts of a running system.
But sadly, the most important ones considered by possible users. The “first sight effect” is important here. So any OS, no matter how powerful it is, will not appeal if the installer isn’t a pretty conglomerate of GUI dialogs, in sequential manner, with defaults tailored for desktop use. Even among “geeks” (those who actually install an OS, because ordinary users aren’t aware of what it is) this seems to be more and more important, and mostly found in… Linux reviews, yes.
If you are talking about home users, I agree with you.
In the enterprise world there are plenty of ugly OS to chose from. For example, I fail to see what systems still running VMS or OS/400 (now z/OS) have to do with pretty.
That’s what I wanted to point out. Home users are considered the primary audience. Server OS like Solaris seem to be more and more regarded a niche market…
You’re confusing terminology: OS/400 has been called i5/OS, and today IBM i; let’s see what name it will have tomorrow, z/OS is what MVS has been (the OS/360 line).
But you’re right: Those systems aren’t very pretty from a home user’s point of view. Still they have strengths (like compatibility back to the 1970’s), and many critical businesses (banking anyone?) rely on them. Also OS/400 is still very popular, but nobody really knows that.
VMS may be an exception. I know that MAN DIESEL did /at least 3 years ago) run VMS software on a SimH-emulated VAX system: http://www.openvms.org/stories.php?story=09/09/25/8154763
In my experience, it’s not that much about “pretty vs. ugly”, it’s about “being different while seeing similarities”. Still, most advantages come for the price of some inconvenience. I know many of them (having used mainframe, midrange and commercial UNIXes in many different forms) and I agree: There is lots of uglyness in them if you look close enough. 🙂
Yeah, I should have looked it up before posting, as it was long time ago I have used such system.
One of my duties during a summer training was to perform backups of an OS/400 system, back in 1994.
OS/400 is not now z/OS, they are different Operating Systems. OS/400 became “IBM i”, z/OS used to be OS/390.
This post will expire of course when IBM decide to rename their Operating Systems again.
This is true. However, more to the point, if the installer doesn’t work then it doesn’t matter how great the rest of the OS is. I’ve seen some pretty interesting projects which simply wouldn’t install on my hardware. At makes the installer a pretty important part of any review: how flexible it is, how it handles errors, whether it has sane defaults, etc.
Is there really anything else that distinguishes one Linux “distro” from another?
The only download links I see on their download page are for Solaris 11 11/11. So… I take it they put up the announcement before actually uploading/updating their download links?
While it is a great OS I think I’ll keep Gentoo Linux on my old Sun Ultra 2 Enterprise (for now). Never had a problem with Solaris per se, other than that hideous [local] terminal display.
A 15″ notebook is probably asking too much, but it would be nice if Oracle offered a single processor tower model, or at least a standard ATX motherboard with processor, so we could build our own system. Even if people did not use Solaris, Oracle would still get a hardware sale.
Engineering those systems is not free, it costs money. The market size for said systems is not large enough for Oracle to bother, since the number of orders would likely not offset the production/design costs. Oracle is in business in order to make money, and they seem to be exceedingly good at it.
Sun used to have them.
At least in the US, you could get Toshiba laptops with Solaris pre-installed.
I remember seeing them on the online shop.
And Sun was so successful at making money… that they got swallowed up by Oracle. You just proved the parent’s point quite handily 🙂
Maybe, but the reasons for Sun’s commercial failure are much more than trying to sell laptops with Solaris pre-installed.
I disagree, the reason for SUN’s demise is very simple: not enough customers were buying (or likely to) enough products for enough money to justify SUN’s cost structures in the medium/long run, so investors bailed. Investors are in it to make money, if a corporation is likely to have decreasing profits then they start to look elsewhere. Personally, I think it is a sociopathic system, and very limited, but that is how things are set up currently.
BTW, the fact that Toshiba may have sold laptops with Solaris preloaded is irrelevant to the point I was making: systems engineering is not free. For Oracle to sell a SPARC ATX board, it would only make sense if the market was large enough to justify the engineering costs for the product. Maybe Oracle could create a marketing campaign to increase the size of said market, but it is clear that has little probability of succeeding given an off the shelf x86 price/performance ratio.
Edited 2012-10-07 01:02 UTC
Ok, there I agree with you.
I think Tadpole still makes ruggedized notebooks with SPARC chips.
But really, SPARC on the desktop has been dead for a while.
I’m shocked to find no mention of a “Metro” style UI…how will this be able to compete against Windows Server?
The reasons of Sun’s commercial failure were trying to sell Solaris and SPARC servers in a world of cheap Linux and x86-64 servers. Oracle might find themselves into the same trouble if they don’t make Solaris have that “something” that will set it apart from Linux, or don’t make SPARC price-performance competitive. Of course they could also try to lock their database software to Solaris, but the backlash will be huge so they ‘ll probably not do it
Desktops and laptops running Solaris, and SPARC desktops and laptops running Solaris, were unimportant pet projects since the early 2000s. They had no significant impact in Sun’s bottom line whatsoever. The age of the RISC workstation ended with the 90s.
Browser: Mozilla/5.0 (Linux; U; Android 2.3.4; el-gr; LG-P990 Build/GRJ23) AppleWebKit/533.1 (KHTML, like Gecko) Version/4.0 Mobile Safari/533.1 MMS/LG-Android-MMS-V1.0/1.2
if they were open source (not the eneterprise things, but enough to make a desktop).Illumos seems a better bet.
DTrace is next to useless without access to the source, so open-source makes sense even in enterprise environments. It’s a shame Oracle doesn’t see that, but then, trying to educate Oracle about the value of community cooperation is like trying to talk to a lawnmower.