Mass Speculation Suggests Oracle May Kill OpenSolaris

CWmike writes to point out that Steven J. Vaughan-Nichols is one of many people questioning where Oracle may land once the acquisition of Sun is complete. One concern that I have heard many people express is that there may be a good chance of OpenSolaris getting the axe for not fitting in with the overall corporate vision. "People outside of IT seldom think of Oracle as a Linux company, but it is. Not only does Oracle encourage its customers to use its own house-brand clone of RHEL (Red Hat Enterprise Linux), Oracle Unbreakable Linux, Oracle has long used Linux internally both on its servers and on some of its desktops. So, what does a Linux company like Oracle wants to do with its newly purchased Sun's open-source operating system, OpenSolaris? The answer appears to be: 'Nothing.' Sun, Oracle and third-party sources are telling me that OpenSolaris developers are afraid that they'll be either moved over to working on Linux or let go once the Sun/Oracle merger is completed."

Comments

Solaris

[Anonymous Coward]

is regular solaris going to live?

Re:Solaris

[Hognoxious]

is regular solaris going to come back to life?

FTFY

Already Open

[Major Blud]

It would be kinda hard to kill since the code is already "open" and out in the wild. Oracle can't prevent the current code base from being forked.

Re:Already Open

[gomek-ramek]

The question, though, is whether a fork would be successful. Without the Sun-paid developers, would OpenSolaris keep its development momentum? My guess is that it would not.

Re:Already Open

[MightyMartian]

I agree. There isn't enough to set it apart from its competitors for it to survive without Sun's active support. I think OpenSolaris is dead dead dead.

I'm wagering it isn't the only Sun offering that's going to be given the boot either. I have a real suspicion that they'll cut OO.org loose too.

Re:Already Open

[mewrei]

Maybe we'll all get lucky and Oracle will just GPL the OpenSolaris code and merge some of the more useful stuff like DTrace and (dare I say it) ZFS into the Linux kernel.

Re:Already Open

[RLiegh]

They can cut off the project's oxygen pretty easily, actually. Most of the project's ecosystem consists of sun-sponsored resources (websites, source code repositories, they also host the mailing lists) and since Oracle will be purchasing Sun's rights they can easily revoke the rights to the binary-only blobs that are required to build a complete and bootable copy of the source tree (if you can't build it, you can't run it -can you?).

Oracle is in a great posistion to kill off Solaris. Considerting that there's little interest or expertise outside of Sun that would be able (much less willing) to maintain Solaris all they'd really have to do is shut down the projects resources and assign the people working on solaris to other projects.

It's sad that Solaris never really took off as a truly free Unix, but it didn't, and now it never will.

Re:Already Open

[DragonWriter]

It would be kinda hard to kill since the code is already "open" and out in the wild.

That makes it hard for them to stop anyone who has the resources and desire from starting their own product based on the OpenSolaris code, but it doesn't make it that hard for them to kill OpenSolaris as an actively developed Sun project.

Not, I should hasten to add, that I think they will do that, just that they can. And if they did, I doubt there'd be a big community keeping OpenSolaris alive after they did. It might survive, but it would become obscure compared to its current status.

Re:Already Open

[elgaard]

And why would Oracle want to prevent if from being forked?

But maybe they could release it also under GPL.

Re:Already Open

[Anonymous Coward]

Or you could just move to FreeBSD.

Re:Already Open

[Bandman]

At the very least, it would diverge far enough from Solaris to be an almost entirely different product.

OpenSolaris....OpenBSD...

Maybe the next version could be called NetSolaris. We could install it on very large toasters.

Re:Already Open

[ocularDeathRay]

other than an enterprise situation where they are already running solaris, is there any reason for anybody to roll out new solaris installs? I am not trying to be a jerk, I really don't know the answer. Is there any essential function that solaris can do, but bsd or linux or whatever cannot do? Not only that, but is there anything that linux and or bsd can't do better, and with a larger community to support it? I could be wrong, but I know for my personal tinkerings I haven't heard anything exciting to me personally about solaris in 10 years. People need to understand that technology evolves quickly, and no matter how emotionally attached we are to an open source project, there is still a good chance that it will become irrelevant at some point.

Don't misunderstand, I am not trying to mitigate the huge pain in the ass this would be to people who are maintaining current solaris systems, but the hard truth is that if volunteers lose interest and there is no big corporate money backing a project, it will likely fall by the wayside. It becomes impossible to even fix security problems if there is not enough developers.

Re:Already Open

[Anonymous Coward]

Somewhere, Ben Rockwood is crying silently in a corner...

Re:Already Open

[MadFarmAnimalz]

Or you could just move to FreeBSD.

I would, but it keeps dying. Don't believe me? Ask Netcraft.

Re:Already Open

[diegocgteleline.es]

Without the Sun-paid developers, would OpenSolaris keep its development momentum?

Another similar question is: Even with the Sun-paid developers, can OpenSolaris keep its development momentum? I very much doubt it, in fact if you look at the trends, you could say that solaris lost that momentum years ago. The only thing that keeps the interest in opensolaris today is ZFS (which is great, but it doesn't make the traditional filesystems irrelevant - LVM and traditional raid suck, but it works and it can do almost everything that ZFS does, even if its a bit slower and crappier), and it's impossible to release big innovative features like ZFS every few years, things like zfs only happen one time every n-decades.

My take: Ellison is not going to follow the anti-Linux competitive attitude that the old Sun had. Its clear that Linux is here to stay, and Oracle couldn't win a fight against Linux, because pretty much everyone except Microsoft and Apple back it. I can't guess what they will do with opensolaris, but it's clear that they aren't going to start a war against Linux, because that would mean starting a war against the huge and increasing share of their Oracle Linux customers.

Re:Already Open

[elashish14]

Also, you can't ignore the issue of where the workers are going to go. When companies restructure, many employees end up getting the axe.

Re:Already Open

[ttldkns]

haven't heard anything exciting to me personally about solaris in 10 years.

ZFS, Nuff Said

Re:Already Open

[Anonymous Coward]

ZFS can run on BSD and Linux (according to Wikipedia)

Re:Already Open

[Midnight Thunder]

I would, but it keeps dying. Don't believe me? Ask Netcraft.

Hmm, surely by the time you reach undead status, you have reached a kind of immortality? So if that is the case, maybe it is the only operating system we should use?

Okay, I will stop trying to over think this.

Re:Already Open

[nurb432]

True, but unless you have the powerhouse ( with a vested interest ) like sun working on it, it might as well be dead as it will stagnate.

Re:Already Open

[BitZtream]

Yes, but thanks the Linux's retarded licensing, ZFS sucks ass on Linux.

Re:Already Open

[tenco]

Maybe the next version could be called NetSolaris. We could install it on very large toasters.

I always wondered if they would run Linux. But you cleared it all up now. It's just so obvious.

Re:Already Open

[ajs]

Maybe we'll all get lucky and Oracle will just GPL the OpenSolaris code and merge some of the more useful stuff like DTrace and (dare I say it) ZFS into the Linux kernel.

That's almost certainly the way they'll go. They already have their own Linux distribution, so there'd be no reason not to do this unless they felt that they were going to derive some value from these technologies being proprietary... again, can't see why.

Re:Already Open

[harlows_monkeys]

It would be kinda hard to kill since the code is already "open" and out in the wild. Oracle can't prevent the current code base from being forked

The notion that once you make something open source, you can't revoke that, is interesting. It's widely believed, but I've seen very little legal analysis to support that belief. What little I've seen from open source lawyers has said that it might NOT be true. I'd love to see a test case.

Some of the factors that would affect a particular case are whether or not the open source license involved is a contract or a bare license. Bare licenses ARE revokable at will by the licensor. In Rosen's book on open source licensing, that is one of the reasons he recommends against using them, in favor of making sure your license is a contract. This is interesting, because one rather prominent open source license, GPL, is not a contract, according to its authors. They are quite insistent about that.

If a particular open source license IS a contract, then whether it is revocable or not will depend on the terms of the contract. Even then, it may be possible to revoke it, if the licensor is willing to suffer a penalty for breach of contract. Contract penalties are almost always just monetary damages, not an order of specific performance. I'll leave it to others to speculate how that would work out.

Another issue is sublicensing. With some open source licenses, if you give me your software, I get my license from you. If I then give the software to a third person, they get their license from me. With other open source licenses, the third person gets their license from you, rather than getting a sublicense from me. GPLv3 is one of the latter kinds of license--it has a specific statement in the license that you cannot sublicense it.

For licenses that are not sublicensible, what happens if the original licensor simply announces that they are giving out no new licenses? People who have the software could still distribute it, free of risk of copyright suit, since they have a license to distribute. But the recipients would not have a license, so they could not redistribute. It might take a way to kill off some open code this way, because it could take a while for all the current owners of copies to stop distributing, but those would probably eventually go away.

Note that I am NOT saying that open source licenses ARE revokable. Just that no one has given a convincing reason that they are not, and that almost nothing else in contract/licensing law is irrevocable, so the notion that open source licenses are irrevocable should be treated with skepticism at this point.

Re:Already Open

[Anonymous Coward]

Or you could just move to FreeBSD.

DTrace and ZFS are both primarily developed by Sun; FreeBSD just imported them.

While DTrace is basically feature complete, ZFS has a lot of stuff that would be useful to have. Built-in encryption is being worked on, and deduplication is rumoured to be coming along. One thing that isn't yet available is removing of devices (and booting off of ZFS is also fairly recent).

It's certainly possible that some company will pick up the Solaris developers and have them code this stuff up for FreeBSD, but it's unlikely.

Re:Already Open

[nicolas.kassis]

retarded eh? It's not like sun was unable to cross license it. They did it on purpose.

Re:Already Open

[tyrione]

dtrace

Re:Already Open

[jvillain]

Ask the dip shit behind proprietary ssh what happens when you open source and then try to take it back. It ain't pretty.

Re:Already Open

[tyen]

Only in user space on Linux, and on BSD some features (integrated iSCSI support for us) that are critical to some sites are missing. We just deployed a new Solaris (paid for the basic subscription support service to get the patches) server to run an inexpensive JBOD disk array that can expand to 384TB of raw disk space using 1TB drives, and ZFS on a paid-for Solaris was the only way to make that project come together on reliability, value, and performance. It is backed by an LTO4 tape library. I treat OpenSolaris as the rough equivalent to RedHat's Fedora; for certain key pieces of infrastructure, there is no substitute for paying up and getting the right technology to get the job done right. I don't see Oracle dumping Solaris, but I wouldn't be terribly put out if Oracle stopped active development for OpenSolaris, and only kept pushing regular updates from upstream with Solaris down to OpenSolaris. Now, if Oracle stopped supporting ZFS, I'd be miffed, but we would migrate over to a LVM and ext4 and live with that.

Re:Already Open

[Anonymous Coward]

I looked into the bsd release
it seems to have performance issues related to ram and other strange quirks... I wouldn't run it on bsd if i had the choice between solaris and bsd ...

Re:Already Open

[Korin43]

It seems incredibly likely that Oracle's fancy Linux will soon have full support for ZFS..

Re:Already Open

[Anonymous Coward]

Which when tested on real applications turned out to be a pile of crap.
The only people excited by this are fat linux devs that would like to steal the code but can't merge Sun code into the GPL-licensed kernawl(as if they cared when they merge proprietary drivers).

Re:Already Open

[Korin43]

It seems pointless to actively kill the project, it would just make the current Solaris users even more annoyed about it. Instead, it would make more sense for them to just stop supporting it and let it die on its own.

Re:Already Open

[rubycodez]

yes, but that's being ported to other open OS. ZFS may long outlive Solaris

Re:Already Open

[ducomputergeek]

yeah, according to them FreeBSD always seems to run about 30 - 40% of their top 10 uptime list month after month.

Re:Already Open

[Anonymous Coward]

That's because it's dead and there's no more patches or upgrades for it. No need to reboot. Just limp along until the hardware dies, then replace it with an OS that's actually maintained.

Those 30-40% of systems are on life support, just waiting to die.

Re:Already Open

[bursch-X]

Wouldn't that be a great slogan? Now with UN*X becoming 40 and everything:

BSD - dying since 1969

or

BSD - 40 years and still dying

Re:Already Open

[nicodoggie]

If you want a zombie computer, you should just install windows.

Re:Already Open

[Kymermosst]

I'll throw in project-based resource controls, zones/containers, trusted Solaris, consistent public kernel API for drivers that won't automatically GPL your code, and a few other things.

Other OSes may offer something similar, but Solaris offers the whole set and largely does it right.

Re:Already Open

[Ilgaz]

Isn't that some kind of justice that BSD in one form reached 10% Desktop use share as OS X?

If we listened to what people said, iPod would never take off, Apple would release 5-10 tablets and go out of business like 5 times, it would be year of Linux on desktop, Microsoft would be dead like 10 times etc.

Oracle is a OS company and produces Linux? I have gave up reading after it. Yes, I didn't read submission even and proud of it. Just like I just laughed at ''Java is doomed'' junk.

Re:Already Open

[Anonymous Coward]

So then the big question is will they do the decent thing and cut zfs licensing loose so the broader linux community can use it with having to screw around with fuse or kernel kludges? Just having the transparent compression on a filesystem natively supported under Linux would greatly help my work. Sure we can store the data on an OpenSolaris box once the next budget cycle open up, assuming budgets hold. But the easiest box to support is the one you don't have to maintain.

Re:Already Open

[Anonymous Coward]

Ok, some valid points here, but let us all remember that for the GPL/LGPL none of the above concerns are legitimate. The author is talking about open source licensing in general.

Re:Already Open

[Fred_A]

haven't heard anything exciting to me personally about solaris in 10 years.

ZFS, Nuff Said

I don't recall nuff saying anything of the sort.
Or anything else for that matter.

Re:Already Open

[harlows_monkeys]

Ok, some valid points here, but let us all remember that for the GPL/LGPL none of the above concerns are legitimate. The author is talking about open source licensing in general

Why would they not be concerns for GPL? It's authors intend it to NOT be a contract, making it a bare copyright license, which is inherently revocable, and GPLv3 explicitly forbids sublicensing. I doubt that FSF would ever try to revoke the GPL on code they own, but there are a lot of projects that have been distributed under GPL where the FSF does not own the copyright.

Re:Already Open

[SpinyNorman]

Why would they not be concerns for GPL?

Because the GPL itself explicitly states that it's non-revocable (unless you violate it).

2. Basic Permissions.

All rights granted under this License are granted for the term of copyright on the Program, and are irrevocable provided the stated conditions are met.

http://www.gnu.org/copyleft/gpl.html

Re:Already Open

[jonbryce]

I thought they had, but it is GPL3 which is not compatible with the GPL2 licence Linux uses.

Re:Already Open

[Anonymous Coward]

Even if an open source license is revoked, the functionality can easily be rebuilt in a new project, because the code is known and can serve as an example for a new build.

Re:Already Open

[SolitaryMan]

This is very sad...

Since Ian Murdock joined the project, I really hoped OpenSolaris would be the "next Ubuntu", but with another kernel.

Re:Already Open

[WaywardGeek]

The Gnome screen reader - Orca - is developed by Sun for OpenSolaris. For me, and others facing visual impairment, this is a biggie. Hopefully, if OpenSolaris does fall by the wayside, Oracle will keeps some of the gems like Orca, and port them to Oracle's version of Linux.

However, this is all very scary. Oracle has not been known as a good open-source contributor, like Sun or IBM. Now they own mysql, Orca, Java, and a ton of key open-source technologies, not to mention OpenSolaris.

Anyone else scared?

Re:Already Open

[mR.bRiGhTsId3]

I seem to remember Solaris being able to outperform Linux on certain workloads, but I can't actually remember which ones. Also, I think the current solaris is still ABI compatible with versions released in the early 90s which could be beneficial to some people.

Re:Already Open

[mehemiah]

um, Unlike linux companies, they can sink, a butload of cash into Networking R&D and come up with this.

Re:Already Open

[Anonymous Coward]

Have you actually looked at the feature list on Solaris between 9 and 10? ZFS has not been the only major improvement. There has been an entirely new TCP stack, zone based virtualization, linux binary compatibilty via system call wrapping in "branded" zones, dtrace, etc. . . Looking forward, opensolaris has introduced crossbow which provides a virtualized network stack which has no comparison in linux.

I'm not going to claim that any one of these features is necessarily as major a change as ZFS, nor that any one of them is reason alone to have interest in Solaris. What I am saying is that I find it difficult to look at the development between solaris 9 and opensolaris and see a lack of "momentum".
 

Re:Already Open

[DragonWriter]

Because the GPL itself explicitly states that it's non-revocable

The GPL is not a law, it is a gratuitous ("bare") license. The law is that gratuitous licenses, whatever their terms, are revocable at will. Ergo, a term in such a license stating that it is non-revocable does not actually make the license non-revocable, since the license does not have the power to change the law.

It is merely a gratuitous promise not to revoke it, which may be enforced as a bar to enforcement of the rights of the copyright holder, to the extent necessary to prevent injustice, where there has been reasonable, detrimental reliance on the promise, by promissory estoppel. But note that there are a whole lot of conditions there, it is not like a contract where enforcement would be far more secure.

Re:Already Open

[lewiscr]

If Oracle stops supporting ZFS, I'd pay for the Veritas Storage Foundation. The VxVM (Veritas Volume Manager) is quite a bit more mature than LVM2, and VxFS is just amazingy fast. It has many advanced features (ie, extra cost), like multi-master cluster mountable FS and Oracle features. Even today, it has a lot of features that ZFS doesn't have. I expect ZFS to catch up, assuming Oracle doesn't screw it up.

Admittably, I have not compared VxFS to ext4. I have compared it to ext3, and there's no way I'll run ext3 on a volume larger than 100G again. fscks are painful

Re:Already Open

[croddy]

I heard they will be renaming the project to Orcale.

Re:Already Open

[dave87656]

The question is, would some other company with enough clout take Open Solaris under it's wings. From what I've heard, it does have a very good kernel for high volume transactional environments.

Complete rubbish

[saleenS281]

Oracle aligned with the Linux project because they could have a say in the direction the OS went, and put back code to the project that they wanted/needed for the wares they were selling to be successful.

Now that they own an entire OS stack, they have no need. If nothing else, I expect unbreakable Linux to fade away rather quickly once the acquisition is complete, as well as Oracle shifting the focus of all future DB enhancements to have a Solaris focus with Linux as a secondary, as was the case historically.

Re:Complete rubbish

[gomek-ramek]

While this is a good observation, I believe there is an argument against this. Oracle has a wide variety of supported platforms. Wouldn't it be more cost-effective to focus development efforts on Linux rather than keeping Solaris alive? They are going to have to maintain Linux either way.

Re:Complete rubbish

[saleenS281]

They can't control Linux, and they know that. Larry wants to own the entire stack, and he's made that very clear for a very long time. There's some choice quotes out there to support that initiative, unfortunately I appear to fail at finding them.

Re:Complete rubbish

[drummerboybac]

Solaris needs to exist to support the Sun SPARC64 and UltraSPARC T2+ processors, the latter of which is a multithreading whiz. It is used extensively where I work, and I hope they keep making it, as 128 simultaneous hardware threads in a 1U can be some powerful stuff when programmed for appropriately.

Re:Complete rubbish

[Bert64]

Not really, they could quite easily adapt Linux to support those processors, it already runs on Sparc and supports the T1 at the very least, i wouldn't be surprised if it already supported the T2+... And the SPARC64 processors these days are made by Fujitsu anyway..

Re:Complete rubbish

[davecb]

There's a good proof of concept Linux that's run on a T2000, but how many years, how many staff and how many debates on LKML would it take to get from a POC to something you could bet your company on?

Honorable bird in hand beats however many in the South Atlantic (;-))

--dave

Re:Complete rubbish

[drummerboybac]

Big difference between running and optimized though.

Re:Complete rubbish

[Anonymous Coward]

Oh my god!! It has started already!! Sun's legendary inability to make up its mind on Linux will infect Oracle. We'll start hearing that Linux is evil, and then Linux is the saviour. Maybe it will be like Solaris/X86 too. Praise it. Kill it. Resurrect it.

We're planning rollouts of RHEL and OpenSolaris/Solaris is nowhere on the horizon.

As a former Sun admin I appreciated SunOS/Solaris because it paid for my first house. Alas, they dropped the ball when it came to small and fast and cheap (funny, considering how Sun was once the small/fast/cheap alternative). They got niched between the really high end and the low end.

I could see Unbreakable Linux going away, but I don't see a benefit to replacing it with Solaris. There are so many deployments happening on x86/x64 Linux that it would be like shooting themselves in the foot.. But hey, Sun was very good at that.

Re:Complete rubbish

[fm6]

It's nice to own your own OS stack. It's nicer to offer what you customers want. Sun owned the same stack and they still had to offer Linux support, because it would have hurt their x64 sales big time if they hadn't. Management will have changed, but not the needs of customers. If anything, there will be a stronger emphasis on Linux, because management will lose a lot of its Solaris-uber-alles bigotry.

All these prognostications about Sun under Oracle are ridiculous. They're all made by people who don't know the first thing about the computer systems business. We start out with people assuming that Oracle will shut down Sun's hardware business "because Oracle is a software company." Now it's a lot of bugs either-or logic about OS choices.

Come to think of it, all of the prognostications people make when Oracle makes an acquisition end up being pretty lame. They're usually based on lame assumptions, like "oh, this acquisition also does databases, they must be buying it in order to shut it down." Which never turns out to be the case.

Re:Complete rubbish

[argosreality]

While they could add in support for the processors, Linux support for sparc (atleast when I played with debian on an older sun sparc based machine) was pretty poor. Somehow I dont see the architecture being fully supported or as scalable as it is under Solaris. Also, how good IS linux at dealing with 128 threads on the fly in one go? I know Linux is used fairly heavily in supercomputers but thats because its cheap and customizable and generally scaled OUT...not up like the T1/2 does

Re:Complete rubbish

[saleenS281]

Not to mention SPARC is the cash cow for Oracle. All the Linux deployments in the world STILL do not equal the amount of money Oracle makes on their legacy SPARC licensing.

Re:Complete rubbish

[Anonymous Coward]

When Oracle has the specs of the processors it would be considerably easier to not have to reverse engineer it. You could have proper support in less than a week.

Re:Complete rubbish

[Anonymous Coward]

Sparc64 is supported plenty well, it's Sun4(x) architecture that isn't supported well. Anything SPARC32 at this point is considered dead, pretty much across the board (kernel, gcc, glibc, etc) since despite the decent amount of hardware still out there, there apparently hasn't been anyone with both the knowledge and motivation to continue fixing bugs on it. The latest STABLE and SMP capable kernel I used on it was 2.2.19 from a debian distro, and as I remember it, while you could get 2.4 to boot on it in SP mode, 2.6 hung midway through the boot (nevermind making sure your kernel fit under the 1 meg limit SILO imposed.)

I've still got a stack of them sitting around, but unfortunately between hard disks, and the sole monitor that can sync with them, they're a PITA to keep running.

Re:Complete rubbish

[Fred_A]

Solaris needs to exist to support the Sun SPARC64 and UltraSPARC T2+ processors, the latter of which is a multithreading whiz. It is used extensively where I work, and I hope they keep making it, as 128 simultaneous hardware threads in a 1U can be some powerful stuff when programmed for appropriately.

Which is why it's such a niche application. Even on a random consumer multi core CPU it's depressing how many processes seem to spawn everything on the same core (more so on an i7 shown with more cores than it really has by Linux).

Re:Complete rubbish

[argosreality]

Umm, the spec and SOURCE for the design of the processors is already opensourced. http://www.opensparc.net/

Re:Complete rubbish

[upuv]

Solaris is not need for UltraSPARC. That's a myth.

UltraSPARC needs Solaris to live on is the more likely scenario.

In the next 12-18 months the we are going to see a fair amount of cheap x86/x64 multicore come on line. I mean more than 8 cores.

For the most part this is going to obsolete the SPARC. Since SPARC development has slowed to a narcoleptic snails pace.

I personally love working with Solaris. It's consistent and strong. Linux is still not up to the Enterprise levels that Solaris is. But man Linux has made massive progress in the last 24 months.

In 36 months those Uber T2 SPARC machines are going in the tip. Sure there is going to be some lost tech when the SPARC dies. There is lots x86/x64 has to catch up on. But one thing it doesn't have to catch up on is price. I can solve the performance gap fairly quickly by tossing another Dell blade in the rack and still be cheaper.

RedHat and Conical(Ubuntu) both are excellent contenders to usurp Solaris as the Enterprise OS in the server room. ( It could be argued that RedHat already has. )

P.S. I won't miss Solaris package management give me a debian packaging any day.

Re:Complete rubbish

[TheRaven64]

SPARC32 MMU support was always painfully broken on Linux, even with the older kernels. We moved some old SPARCStations from Linux to NetBSD and the speed increase was visible to the user.

Re:Complete rubbish

[TheRaven64]

I know Linux is used fairly heavily in supercomputers

This is true, but is incredibly misleading. If you look at something like a big SGI machine with, say, 512 processors running Linux, it's really a number of nodes with 4 processors, each with one its own Linux instance. There is some custom hardware for the fast interconnect and distributed shared memory. The kernel is slightly modified to support process migration between nodes.

A lot of the other 'Linux' supercomputers (for example, IBM offerings) use Linux for job control and I/O, but use a custom, lightweight OS for running the real tasks.

In contrast, Solaris has been running on SMP machines with 128 or more processors for a decade. Recently, the entire TCP/IP stack was rewritten to be a set of small independent components so it, and a number of other kernel subsystems, scale to a large number of processors. Solaris also has better support for self-healing systems, ZFS, DTrace (which is useful for Oracle because it helps tune the database) and a much mature codebase. Another advantage Solaris has is the ability to plug schedulers. Linux recently replaced the scheduler with a brand new one which has worse performance characteristics on some workloads, better on others. This kind of change is exactly what Oracle customers hate. With Solaris, a new scheduler can be enabled or disabled on a specific install.

Re:Complete rubbish

[davecb]

The chips are open-source. Oracle doesn't need to do anything.

--dave

Re:Complete rubbish

[Bert64]

The MMU on the Sparcv7 cpus was horribly broken on Linux... But to have a CPU like that, you need something like a Sparcstation 2 or similar. I had Linux on an SS2 and also migrated it to NetBSD for the same reason you mentioned.

I used to run Linux on a Sparcstation 20, which had 4 Sparcv8 cpus and it ran pretty well... It definitely ran a 2.4.x kernel perfectly well, not sure if i put a 2.6.x on it. I still have the box, and the disks somewhere - it used to be my mail server for many years, and i only replaced it because spam filtering was starting to tax it too much.

I never used them with a monitor, because serial console is far more useful..

Incidentally, netbooting gets round the silo limitations if you need to boot a big kernel, or just compress your kernel and use modules for everything else... I always ran stripped down kernels anyway so it wasn't an issue.

Re:Root is like crack

[Anonymous Coward]

This would have been funny as an XKCD comic.

Don't believe it....

[GuyverDH]

opensolaris - the regular SXCE builds are Sun's testbed for new updates, patches, fixes and technology updates...

It's noted as 5.11 for the version, codenamed Nevada.

It's very similar to the way the unix kernel builds happened at one time (to be honest I haven't looked lately to know if they still do this or not) - in that the even number release is production and the odd numbered release is development...

Unless Oracle intends to kill off Solaris altogether, I don't see them killing OpenSolaris.

Re:Don't believe it....

[GPLDAN]

I agree here. The exact opposite may be true. Unbreakable Linux will become Unbreakable UNIX - and it will be increasingly based on OpenSolaris. There are a lot of developments in Solaris - and I think Ellison is perhaps unhappy with his relationship with Linux.

I can tell you as a IT Director in finance that they have come pushing Unbreakable into big accounts, and want to cut Redhat off at the knees as much as possible. So the opening salvos have already been sent, and sinking Redhat and getting all Oracle installations off it is the goal.

Re:Don't believe it....

[Anonymous Coward]

It looks like you've misinterpreted the Solaris version numbering scheme. 5.10 is the current Solaris 10 range, which was preceded by 5.9 (Solaris 9). 5.11 is most likely Solaris 11. My current Solaris box, for example is running Solaris 10, update 6, and looks like this:

-bash-3.2$ cat /etc/release
                                            Solaris 10 10/08 s10s_u6wos_07b SPARC
                      Copyright 2008 Sun Microsystems, Inc. All Rights Reserved.
                                                Use is subject to license terms.
                                                        Assembled 27 October 2008
-bash-3.2$ uname -a
SunOS logan 5.10 Generic_139555-08 sun4u sparc SUNW,Sun-Blade-1000
-bash-3.2$

Re:Don't believe it....

[GuyverDH]

There's no mis-interpretation.

If you would care to read, and get an understanding, please try the following link..

http://whacked.net/2005/06/21/confused-so-was-i/

Solaris 11, will continue to be the development cycle, Solaris 10 is (as put by one of the developers) *THE LAST VERSION of SOLARIS EVER*... They may achieve Solaris 10, Update 535 in time, but as of this moment, Solaris 10 is the highest production version we'll see.

So Nevada 5.11 build xxx are all development releases.

Re:Don't believe it....

[afidel]

Man, them trying to cut Redhat off at the knees is so short sighted because
A) their support organization sucks (maybe this will get better if they use the Sun side of the house to answer non-DB questions)
B) Redhat employs a bunch of people that work on linux without costing Oracle a dime.

Re:Don't believe it....

[gmack]

Actually Oracle also employs some high level kernel developers. A lot of the FS work that has been done lately has been done by people with Oracle email addresses.

Just wondering

[JonJ]

Has Nichols ever hit a homerun on his speculations? Most of the time, he seems to me like an old man that just can't seem to connect the dots.

time to steal features

[bcrowell]

For anyone already committed to OpenSolaris, there are some obvious things to do: (1) Celebrate the fact that it's open-source, which limits how badly you can be screwed. (2) Write a plan to start transitioning to Linux or FreeBSD or whatever. (3) Help to organize a community operating outside of Oracle that will coordinate on maintaining the OS with security patches for the rest of its lifetime.

For anyone else, now would be a good time to think about stealing features. I know a lot of people really like DTrace. Well, it's already been ported to FreeBSD, and the Linux port seems to be nearing completion.

Re:time to steal features

[MightyMartian]

How much penetration did OpenSolaris ever achieve? I know a few guys that through it up just to take a peak, but I doubt very much that there are that many production machines out there. It always struck me as more of a curiosity. But I dunno, maybe it's all over the place.

Re:time to steal features

[Anonymous Coward]

I work work for a 40,000+ person company and we have many dozens of production OpenSolaris servers and tons of workstations.

Re:time to steal features

[Tanktalus]

Really, I thought that anyone watching Larry go about his regular routine that the appropriate response might just be to abandon Larry-ville the same way /. seems to advocate abandoning Balmer-ville. But maybe that's just me (running IBM DB2 on Linux...)

Re:time to steal features

[Anonymous Coward]

Why would there be any OpenSolaris deployments?

It was never marketed as a finished, production-ready product. It's a development project, OpenSolaris, when ready will be released as Solaris 11. The current production-ready, stable release is Solaris 10.

Who in the hell deploys a development OS that isn't production-ready in a production environment?

You'll never hear of large OpenSolaris deployment, even after it is ready - You will hear of people deploying Solaris 11, however.

Re:Root is like crack

[$RANDOMLUSER]

Been done.

Okay, you are a sandwich

[Anonymous Coward]

From a late-70s/early-80s Saturday-morning cartoon public service ad touting healthy eating:

Cartoon Kid: "Can you make me a banana?"

Cartoon character touting good eating: *wave of magic wand* "Okay, you're a banana."

Re:Okay, you are a sandwich

[Anonymous Coward]

That was a Burger King commercial, and it was a milkshake and not a banana the kid wanted.

Makes absolutely no sense

[javacowboy]

Why would Oracle kill Solaris? Their first public pronouncement on the Sun takeover specifically mentioned Solaris next to Java as the reasons they want to acquire Sun. Killing Solaris would be almost as much of an about face as killing Java.

Solaris represents one of Oracle's differentiators. It has features that Linux can't due to licensing concerns, namely ZFS and DTrace. It gives them the opportunity to add value to their offerings, as opposed to being simply a reseller, which is what they'd be if they'd favour Linux.

What's more, Oracle's database is well-known to run better on Solaris than on any other operating system. Killing Solaris would remove that competitive advantage.

The only reason Oracle supported Linux so strong is that they didn't have an OS of their own. When they acquire Sun, they will.

Re:Makes absolutely no sense

[MightyMartian]

I don't think anyone doubts that Solaris will go on. But I see little advantage in Oracle's case for continuing to dedicate resources to OpenSolaris.

Re:Makes absolutely no sense

[javacowboy]

In the same vein that StarOffice is built on top of OpenOffice, Solaris is now built on top of OpenSolaris. The next version of Solaris will be cut from a version of OpenSolaris.

Besides, if you RTFA, the author is trying to argue that Oracle will kill all of Solaris, not just OpenSolaris.

Re:Makes absolutely no sense

[jeffstar]

What was the advantage to Sun to have OpenSolaris? Whatever it was, Oracle will likely have the same reason to continue dedicating reasources to OpenSolaris as Sun did.

Re:Makes absolutely no sense

[$RANDOMLUSER]

What's more, Oracle's database is well-known to run better on Solaris than on any other operating system. Killing Solaris would remove that competitive advantage.

Indeed. Also, support for Solaris will be a revenue stream for Oracle as well. Solaris on big-boy hardware in the data-center isn't going anywhere any time soon. However, OpenSolaris only attracts people trying to do it on the cheap. Oracle can move those people to Unbreakable and plug up the money drain that is OpenSolaris.

Re:Makes absolutely no sense

[DragonWriter]

I don't think anyone doubts that Solaris will go on. But I see little advantage in Oracle's case for continuing to dedicate resources to OpenSolaris.

OpenSolaris serves to help promote Solaris, which is why Sun introduced it. There would be little sense in Oracle killing OpenSolaris if they intended to try to continue Solaris as an OS.

Re:Makes absolutely no sense

[UID30]

What you talkin 'bout Willis? Oracle's primary development platform has been Linux for years now. I think the vague "runs better" test is pretty much a wash when you compare optimized code builds running on similarly powered hardware.

I think Sun hardware is really more of a vanity thing in business nowdays ... so "company a" can look down their nose at "company b" and say "we dont use Dell servers, we're a Sun house"...

OMG! THANK you for making me post this! I NOW understand the Oracle-Sun merger! They're both "vanity" business models! Its been bothering me since the merger was announced ... but now I see the synergy plain as daylight. Its all about super large corporate businesses and absurdly high maintenance contracts.

Wow. That is some kind of evil genius. I'm going out to buy some Oracle stock.

Re:Makes absolutely no sense

[Anonymous Coward]

To what extent did it accomplish that goal? That is the question Oracle is going to ask.

Re:Makes absolutely no sense

[argosreality]

The advantage is fairly obvious - its open, so people can play with it early. they can pickup bugs, they can code new works based upon, etc. so Sun uses OpenSolaris as its "open-beta" before rolling changes back in the core, enterprise stable version of Solaris

Re:Makes absolutely no sense

[jabuzz]

At the moment there is only one company I can goto and by a database with support and not have the vendor be able to pass the buck. That company is IBM with the DB2 on AIX on Power stack. It is IBM end to end. If Oracle take over Sun, then there will be another stack in the mix, Oracle on Solaris on Sparc. That has to be some selling point.

That said if I where Larry, I would do away with OpenSolaris, but only because I would make no distinction between Solaris and OpenSolaris. I would push it for all it is worth against Linux

Re:Makes absolutely no sense

[sysgeek01]

I don't have ADD. I was just... OMG a waffle!

Re:Makes absolutely no sense

[bky1701]

If that were true, Oracsun would be planning to merge with Apple, but that could never... OH SHI-

Re:Makes absolutely no sense

[sydb]

You forgot DB2 on iSeries and DB2 on zSeries.

Linux is the biggest fish in the "open" space.

[reporter]

Unless Oracle explicitly spends resources to develop OpenSolaris, it will fade away and die in the "open" space as Linux is the biggest fish there. The typical geek who builds a freeware application builds it for Linux first since Linux is the dominant freeware operating system.

So, what is the chance that Oracle will spend resources on OpenSolaris? The probability is exactly 0.

Oracle -- along with Intel and Cisco -- is notorious for viewing engineers as dots on a graph and rating them on a bell curve, firing the bottom 10% annually. These companies do not waste any money or time on "underperformance" by either engineers or products. If a product does not produce any revenue, then it is abandoned.

This shark-like mentality has gained popularity in recent years among American companies.

Re:Linux is the biggest fish in the "open" space.

[Anonymous Coward]

No - linux is not the "biggest"

If you take the mass of all the things that call themselves "Linux" - then yes.

Linux is a kernel. But RedHat's efforts are different than Debians, which is a different set of efforts than Canonical, which is .....

And Apache, while running on GNU/Linux forks is not Linux. Nor is OpenOffice. Or .....

Re:Linux is the biggest fish in the "open" space.

[kindbud]

Obvious troll is obvious. Well, except to all you n00bs who modded him "interesting."

Re:Linux is the biggest fish in the "open" space.

[Anonymous Coward]

This shark-like mentality has gained popularity in recent years among American companies.

You are mistaken, this has always been the reality side of the "American Dream (tm)", anything that gives you a winning hand is good. Or to put it in videogame terms : your self-defined mental blocks prevent you to win, now get off my ring scrub!. But, may be you would prefer a car analogy?

Re:Linux is the biggest fish in the "open" space.

[thommym]

Two comments. 1. Not only is Linux the biggest fish, it's by far the ugliest too 2. OpenSolaris development wouldn't die if Oracle only layed off the bottom 10%. They'd have to start picking from the positive sigma side...

Re:Linux is the biggest fish in the "open" space.

[Anonymous Coward]

I think UNIX with Solaris is still the biggest fish in the open space.
Mind you that plenty of companies (HP, Sun, Oracle, Cisco, IBM, etc...) have many installations in Unix, Solaris and OSF stuff. Most of them are critical systems too.

Linux just has the biggest mouth in the "open" space to date. Guess what, it's still smaller than app-level stuff like Java, PHP, Python, and Apache stuff). It still has a number of years to go and more hardening to a core kernel solution (vs the forking that everyone is doing...).

Dead??

[DaMattster]

OpenSolaris will not be completely dead. The community at large will pick it up and it will take on a life of its own much in the same way as BSD UNIX was when the Berkeley CSRG group disbanded. OpenSolaris is still important and used heavily throughout industry. It is not my intention to start a flame war, but Solaris is even more mature as a platform than Linux. I am a fan of all open source operating systems and software because it takes computing out of the power of the corporation and puts it in the hands of the users.

Re:Dead??

[Anonymous Coward]

Using the term "heavily" maybe exaggerating a little

Re:Dead??

[Fred_A]

OpenSolaris will not be completely dead.

It depends whether the new Sun management is willing to collaborate...

Oracle : Bring out your dead !
Sun : I've got one. nine pence.
Solaris : I'm not dead!
Oracle : What ?
Sun : Nothing -- here's your nine pence.
Solaris : I'm not dead!
Oracle : Here -- he says he's not dead!
Sun : Yes, he is.
Solaris : I'm getting better!
Sun : No, you're not -- you'll be stone dead in a moment.
[ ... ]

They should spin it off - as a non-profit?

[davidwr]

It's open-sourced, so theoretically anyone can pick it up.

However, to be good corporate citizens, if they don't plan on keeping it they should spin it off and provide enough seed money to let a few employees go with it for a year or so.

One of my favorite quotes...

[UID30]

I had enough exposure to Solaris in the 90s ... I remember when a Sun install team put in the 1st e4500 16 processor high availability box at my employer ... they had powered it up and had a bunch of our company VPs standing around the cold room oogling it ... the Sun rep was giving an executive overview of its HA features, full hot swap of processor boards, power supplies, yadda yadda yadda. My (then) boss, a lowly manager in the VP crowd, walks up to the e4500 and pops a processor card out ... the whole system seg faults an UGLY death. Ahhh ... good times.

If operating systems are weapons, Solaris is a World War II German railway gun with a cracked breech block.
- Charlie Stross

Re:One of my favorite quotes...

[drummerboybac]

the 90's.....that was, like, 10 years ago. Take a look at most OS's 10 years ago and see how much difference there is.

Re:One of my favorite quotes...

[Bert64]

You sure you had 16 cpus?
The E4500 has 8 slots, 2 cpus per slot, but you need to use at least one of those slots for an IO board otherwise you have no scsi and no networking, so the practical limit is 14 cpus...

Re:One of my favorite quotes...

[Anonymous Coward]

Compare how Linux handles the same processor offline event.... ten years ago. :)

Seriously, the E4500, so that was probably Solaris 2.6, which didn't support the RAS features, regardless of what you were told. I have actually removed memory and CPU boards from a running Solaris 8 system (after they had been properly offlined)... on SunFire hardware.

There's so much FUD surrounding Solaris or Linux or Windows by various zealots. I hate you all.

Re:One of my favorite quotes...

[hoggoth]

I had a similar experience when I was at N.E.C. We were showing off one of our fully redundant servers to some execs from a Wall St. firm (I won't name them, but they are still in business, but with a merger). While my manager was talking about how fail-safe the server is one of the execs walked around behind the rack and just jammed his pen through the fan in the back to see what would happen.
Luckily back-up fans spun up and everything was fine, but there were a lot of sweaty foreheads in the room...

Re:One of my favorite quotes...

[UID30]

Not entirely sure, no. Just like now, back then I was a lowly developer ... not fit to bathe in the glory of Sun reps or gawking VPs.

I'm just glad I was close to the right number ... it was a mixed house of e250, e450, e4500 ... i think there were a few e6500s and one or two e10ks floating the data center also.

... and for some reason, with all that hardware, I seem to remember complaining about disk space on a daily basis.

Re:One of my favorite quotes...

[HogGeek]

Maybe they passed all of the I/O through the Two RS-232/423 ports on clock board...

Re:One of my favorite quotes...

[Fred_A]

the Sun rep was giving an executive overview of its HA features, full hot swap of processor boards, power supplies, yadda yadda yadda. My (then) boss, a lowly manager in the VP crowd, walks up to the e4500 and pops a processor card out ... the whole system seg faults an UGLY death. Ahhh ... good times.

Nothing kills a demo like a manager. The rep should have remembered to tie them to their chairs.

Re:One of my favorite quotes...

[mzs]

You know why it panic()ed (no not a segfault)? Because the processor board was not offlined first. That guy was a moron, more over since the boards on those systems did not have pins of sufficiently different lengths it could have fried an expensive component to simply yank the card out.

Look at the bright side -- ZFS for Linux!

[Wrath0fb0b]

I've long been immensely frustrated that you can't get kernel-space ZFS (sorry FUSE) compiled into a Linux kernel because of inane licensing issues*. Someone should write a patch for those of us that want to compile it ourselves on the theory that the FSF would be insane to sue a personal user of open-source software for daring to compile it with other open source software of a different flavor.

* Porting ZFS to Linux is complicated by the fact that the GNU General Public License, which governs the Linux kernel, prohibits linking with code under certain licenses, such as CDDL, the license ZFS is released under. [Wikipedia]

Re:Look at the bright side -- ZFS for Linux!

[DragonWriter]

Porting ZFS to Linux is complicated by the fact that the GNU General Public License, which governs the Linux kernel, prohibits linking with code under certain licenses, such as CDDL, the license ZFS is released under. [Wikipedia]

Nothing in the GPL prohibits linking with code under any other licenses, per se, however, many other licenses do not give one the rights one would need to relicense the code under the terms in the GPL (either instead of the terms in the other license, which is required under the GPLv2 [under which the Linux kernel is licensed], or in addition to the terms of that license, which is an option in certain cases under the GPLv3).

Re:Look at the bright side -- ZFS for Linux!

[Bert64]

Yes, someone should port it and only provide it as source or a diff, there shouldn't be any licensing issues there since it isn't linked yet, and the GPL does not apply to anyone who just compiles it for their own use and doesn't distribute the binaries...

Re:Look at the bright side -- ZFS for Linux!

[The-Pheon]

I've long been immensely frustrated that you can't get kernel-space ZFS (sorry FUSE) compiled into a Linux kernel because of inane licensing issues*....

Well it is a good thing FreeBSD does not have a restrictive license like that. FreeBSD 8.0 will have ZFS with zpool 13, and here is how to use it.

http://wiki.freebsd.org/ZFSQuickStartGuide

Cheers!

Re:Look at the bright side -- ZFS for Linux!

[Anonymous Coward]

Can't you just patch it yourself and use it yourself? I've often wondered about rogue code warriors patching up their own Dr. Frankenstein's Monster Linux/BSD.

Any of you guys exist in reality or just in my head?

Re:Look at the bright side -- ZFS for Linux!

[diegocgteleline.es]

The problem with that is that ZFS is not just a filesystem, it's a complete "IO stack". It's everything that does from the VFS to the device drivers. Sun didn't improve their old stack, they wrote a new brand system and they left the old system there.

Such thing would not be tollerated on the Linux main tree, it would be considered a very ugly design mistake. For them, the IO stack would need to work for ZFS and for FAT, and they would never buy the logic of "ZFS is special and needs special treatment to be better than the rest". If ZFS was released, Linus & co wouldn't accept it until ZFS is modified to fit the Linux IO stack, and/or they modify the Linux I/O layer to fit what ZFS needs.

Re:Look at the bright side -- ZFS for Linux!

[Anonymous Coward]

Which is why it actually fits pretty well with FUSE.

The ZFS-FUSE port is pretty good when you consider it was basically written by one guy during a Google SoC. If it had a dedicated developer team behind it, it might be usable for production loads eventually.

People who suggest 'just write a patch' to put ZFS in Linux don't realize how much work it would be. Even after all that work you'd end up with a legally-questionable, difficult to distribute, suitable for personal use only version of ZFS which would probably be less reliable than the FUSE version. I can't imagine why nobody has stepped up to do it yet!

Re:Look at the bright side -- ZFS for Linux!

[Anonymous Coward]

...compiled into a Linux kernel because of inane licensing issues*. Someone should write a patch for those of us that want to compile it ourselves on the theory that the FSF would be insane to sue a personal user of open-source software

Where someone ( particularly on the Opensolaris side ) to have suggested doing this with a GPL based project the immense outcry would be heard on the moon. Why not re-implement it per the spec ( I know someone who did it in Java ), or just use OpenSolaris.
OpenSolaris NAS, nothing short of amazing.

Re:Look at the bright side -- ZFS for Linux!

[Wrath0fb0b]

People who suggest 'just write a patch' to put ZFS in Linux don't realize how much work it would be. Even after all that work you'd end up with a legally-questionable, difficult to distribute, suitable for personal use only version of ZFS which would probably be less reliable than the FUSE version. I can't imagine why nobody has stepped up to do it yet!

(1) I stand corrected as to the technical difficulties in the task.

(2) Distribution and legality would be easy -- just distribute it as source code + patches to the normal linux build system, which cannot possibly illegal. Let the end-user decide whether he feels that his actions violate the spirit of the GPL (violating the words is conceded, but it's ridiculous on its face that the wording is not flexible enough to allow the end user to basically compile both separately and link them together.)

Re:Look at the bright side -- ZFS for Linux!

[tbuskey]

Yep, provide sources & patches like Minix did before it was open sourced.

Re:Look at the bright side -- ZFS for Linux!

[Just Some Guy]

FreeBSD 7-STABLE already does. I'm running it on several machines with good results.

Re:Look at the bright side -- ZFS for Linux!

[Anonymous Coward]

Right - because that is the fundamental design of ZFS - the filesystem layer works with the data management layer so that they all know what is going on and hence you have consistant copy-on-write operations. To maintain consistancy and performance it does its own IO scheduling - not all that different to UFS but far more advanced. All the cool features that come out derive from this simple integration.

Its not as straight forward and dictating that it should integrate into LVM or provide its interfaces to UFS - they simply dont work that way and ZFS cannot work without the fully integrated stack.

Re:Look at the bright side -- ZFS for Linux!

[Anonymous Coward]

I may be wrong, but I thought the reason FreeBSD had a lot of this stuff is because they include it in source form, don't compile it in stock kernels, then tell you 'if you want this feature compile it in, but it's governed under X license terms'.

Mind you I haven't run FBSD since 4.7-4.8 but I seem to remember something like that when rolling a custom kernel for a server I was running.

Re:Look at the bright side -- ZFS for Linux!

[TheRaven64]

It wasn't in the stock builds because it was not stable. It is now included, but in modules. If you don't load the modules, then you are only running BSD-licensed code. The CDDL and BSDL don't have any restrictions on linking, but some users wanted a pure BSDL kernel. With FreeBSD 7, because ZFS was still very experimental, there were two install CDs provided, one with ZFS and one without. Now there is only one, with ZFS support.

Re:Look at the bright side -- ZFS for Linux!

[Anonymous Coward]

INALB, GPL doesn't prevent end users from linking to anything.
It does prevent distribution of a linked version and the non-GPL entity from bundling their code with GPL stuff.

OTOH, if the library/extension could be easily linked to the Linux kernel, but it just takes extra effort by the users, that's fine.

Need a precedence? nVidia video drivers and MP3 audio codecs. Users have to chose to add these things in.

One word: Dtrace

[seifried]

One thing Linux is lacking (and will possibly never have due to politics) is Dtrace, which is sad because a) Dtrace kicks ass, b) it's mature and works well and c) system tap is... well.. one day when a vendor ships it I guess we'll find out how well it works. This is one spot OpenSolaris and Solaris (and Mac OS X which now has Dtrace) really shine, you can extract useful telemetry and performance data from the system easily.

Re:One word: Dtrace

[bmcage]

The port to linux is ready, it is another licence you can compile in yourself: http://www.crisp.demon.co.uk/blog/archives/2009-06.html#2009-06-28T12_07_05.txt

GPL ZFS

[TyFoN]

Maybe we finally will see GPLd ZFS now even though btrfs is superior in design. I wouldn't really trust ZFS to hold my data given all the ignored bug reports about data corruption.

Old unix'es rarely ever really dies.

[DUdsen]

AIX is still somewhat in support, HP-UX the same no OpenSolaris will be around for decades to come we might see Oracle stop pushing it actively for new customers but you dont kill a prodoct like that not with the price some organisation is willing to pay for sevice and support deals for existing systems.

Sometimes it's not about the strategic game of cat and mouse and all about the cash flow.

Re:Old unix'es rarely ever really dies.

[UID30]

I beg to differ. Case in point: Ultrix. I think it is safe to pronounce DEC Ultrix as D E A D. *grin*

Re:Old unix'es rarely ever really dies.

[jabuzz]

What do you mean AIX is still somewhat in support!!!

It is still in development, with new versions coming along and IBM are still producing new hardware for it. Admittedly none of it is cheap, an entry level p520 express is still eye wateringly expensive if you are used to x86 hardware, but it is just as much an alive platform as it ever was.

Re:Old unix'es rarely ever really dies.

[afidel]

Tru64 is dead, HP has announced that it will no longer be supported past 2012 and that the last maintenance release will come out next year. They even shelved the plans to grab the best parts and graft them into HPUX. IRIX is dead as of 2013. Openserver might as well be dead as no sane IT manager would touch it with a ten foot pole. Basically there is only Solaris, AIX, HPUX and OS X as true Unix and then the unix-alikes linux and BSD derivatives.

Re:Old unix'es rarely ever really dies.

[ci4]

Tru64 is dead, HP has announced that it will no longer be supported past 2012 and that the last maintenance release will come out next year. They even shelved the plans to grab the best parts and graft them into HPUX. IRIX is dead as of 2013.

Another one with a time machine. Where do they sell them?

Could you hush me the Lotto numbers from next Saturday, please?

Openserver might as well be dead as no sane IT manager would touch it with a ten foot pole. Basically there is only Solaris, AIX, HPUX and OS X as true Unix and then the unix-alikes linux and BSD derivatives.

GPL...

[Bert64]

If Oracle don't want to commit resources to developing solaris, they should triple license (including GPL) it... Solaris is too widely used to die, so third parties will continue developing it and having it GPL licensed will allow drivers to flow from linux (which linux has a lot more of and solaris is very much lacking) and zfs/dtrace to flow back.

Perhaps FUD - Complete rubbish

[davecb]

If I wanted to capture business from Sun, I'd start a rumor that Oracle was going to get rid of big parts of Sun.

And, just to add insult to injury, the rumor would have them laying off the people Oracle most wants to retain!

--dave

Dtrace on FreeBSD

[witherstaff]

Don't forget the nice Dtrace on FreeBSD. Great stuff for servers.

This just in...

[Temujin_12]

This just in: "Mass Speculation" also suggests:
1) The world will end in 2012
2) Man never landed on the moon
3) Vaccines cause autism
4) Technology = magic
5) Science is infallible
6) Religion is infallible
7) Windows is better than Mac
8) Mac is better than Windows
9) Mac is better than *nix
10) *nix is better than Mac
11) Windows is better than *nix
12) *nix is better than Windows

I really need to meet this "Mass Speculation" guy. He seems to be all over the board on things.

Re:This just in...

[shis-ka-bob]

I thought this was insightful.

Re:This just in...

[Anonymous Coward]

Anyone who's not an idiot knows the real truth: Plan 9 is better than all those other shitty OSes.

Re:This just in...

[noidentity]

If his Oracle were any good, he wouldn't even need to speculate!

Re:This just in...

[Anonymous Coward]

Usually this sort or 'speculation' only gets to the press because it has 'some' reality on it: SJVN source's certainly are Sun's former developers and project managers involved on Open Solaris... it's much easier to hide a buyout rumour (only top execs involved) than the cancellation of a product line (everybody onboard has to be involved, sooner or later). Once someone who disagrees knows, everyone knows.
Think about how the Bush administration tried to shut the mouth off of the people involved in torture and other criminal activities... Or the Nixon one, if we go even further... As this is business and not politics, we'll certainly know the truth in just a few days, a week or two at most.

Oh, no, Not that!!!

[icebike]

Quote: "OpenSolaris developers are afraid that they'll be either moved over to working on Linux or let go"

Let see, Job and Paycheck working on something with a future, or sulking at home working on a dead end?

Decisions....

Re:Oh, no, Not that!!!

[lems1]

Well said

Solaris internal to Oracle

[panic]

I think most people underestimate how much solaris oracle uses internally...

There is marketing hype.. then reality

OpenSolaris == Fedora

[jregel]

The value of OpenSolaris to Sun is the same as Fedora is to Red Hat Enterprise Linux; it's the cutting edge release that allows the new features to be added without compromising the stable release. It's improving as a desktop operating system, but that's not the real point of OpenSolaris. Solaris is primarily a server operating system and that's where it excels. It manages to include things today such as ZFS and Dtrace that will one-day have equivalents in Linux. These technologies are already mature on Solaris. Code from OpenSolaris is also used by the Sun OpenStorage platform and presumably will be the basis of the Sun OpenNetwork platform.

Before I'm modded down as a Linux-hating, Solaris fan-boi, I'm posting this from my home Linux workstation, sat next to my OpenSolaris server. Sometimes it's about the technology itself and not technology religion.

Re:OpenSolaris == Fedora

[Super_Z]

Code from OpenSolaris is also used by the Sun OpenStorage platform

OpenStorage is basically OpenSolaris plus a kickass web-gui.

Re:OpenSolaris == Fedora

[thatkid_2002]

Crap! Sorry about the troll mod. Totally accidental!

Once you are modded you cannot be unmodded :/

Re:OpenSolaris == Fedora

[_Shad0w_]

Unless the modder post on the thread, like you just did :)

Personally I like Open Solaris

[thetoadwarrior]

I have the latest version, which I've not tried yet, I can't decide which system is less of a hassle to back up and install a new OS on.

It had a few issues upon it's release but it's very nice. Most stuff works I need works with it and it has some nice thins not found elsewhere (ZFS, DTrace). Personally I think it'd be dumb to get rid of it. They should promote it more and hopefully get it to grow and then control what they can't through Linux.

Who cares?

[Anonymous Coward]

The 104 users?

fuck you

[mungtor]

you're probably right. As much as I wanted to find fault and prove you wrong, I can't and now I'm just bitter.

Unlikely -- Oracle optimizes for Solaris 10/SPARC

[crmartin]

Their highest capacity versions and licenses are all for Solaris 10 and SPARC. And, as someone else noted, it would be hard to kill OpenSolaris, because it's already Open. Like MySQL, if they tried to close it, it would just branch (as MySQL already is.)

First of many

[nurb432]

I see most of Sun's work going away the same way. No real business reason for Oracle to keep it. ( and they are just bastards anyway )

No years - Canonical are betting their company

[anti-NAT]

http://www.ubuntu.com/partners/sun

Re:No years - Canonical are betting their company

[davecb]

I know they did the port, and it clearly works, but I'd suggest it is the kind of low-priority thing one does to make sure they have a presence in the machine room, and predominantly on the Sun x86 machines they mentioned in the cited page.

It would be somewhat weird to bet the success of Ubuntu, the user-friendly desktop and laptop Linux, on the success of a port to rack-mount machine-room SPARC server hardware.

--dave

Re:No years - Canonical are betting their company

[davecb]

Seriously, though, I think the article is mostly FUD, and that Oracle won't risk their company on throwing away a working OS.

In fact, if they want to save money, they would encourage more Open Solaris efforts (;-))

--dave

Re:No years - Canonical are betting their company

[Dan Ost]

How would Open Solaris save them money over using Linux?

Re:No years - Canonical are betting their company

[davecb]

Not at all: I was suggesting that they wouldn't want to drop it, but instead encourage more of it.

--dave

Re:No years - Canonical are betting their company

[anti-NAT]

Let me quote your original statement:

"There's a good proof of concept Linux that's run on a T2000, but how many years, how many staff and how many debates on LKML would it take to get from a POC to something you could bet your company on?"

isn't this contrary to what you've now said?

"I know they did the port, and it clearly works, but I'd suggest it is the kind of low-priority thing one does to make sure they have a presence in the machine room, and predominantly on the Sun x86 machines they mentioned in the cited page. "

A low priority port (that they've certified for certain releases), which I wouldn't necessarily disagree with, is a long way from a supposed P.O.C. that nobody would be willing to put any money behind.

Re:No years - Canonical are betting their company

[davecb]

I consider them pretty close to synonymous, and wouldn't recommend Oracle or Canonical bet the company on it (;-))

--dave

"Nothing" would be a fine choice if

[melted]

"Nothing" would be a fine choice if:
1. They GPL2d ZFS, DTrace and other core technologies that make Solaris attractive
2. Allocated the dev resources to port them over (or in the case of ZFS adapt them to the existing kernel better).

If they just rm -rf it, it'll be a very sad decision indeed.

GPL Is A Distribution License!

[JerkBoB]

RAAARGH!
.
Legally, there is no reason in the world that you (or I, or my dog) can not port ZFS to Linux and run it ourselves. The licensing issues only come into effect when you want to DISTRIBUTE your work as binaries. The idea that the FSF would sue someone for linking ZFS and Linux together in their basement is so fundamentally misguided that I would be shocked if this weren't a /. discussion.
.
Let me underscore this point: GPL ONLY COVERS BINARY DISTRIBUTION! The goal is to prevent someone from taking GPL-covered code, compiling it to binary, and then redistributing it without accompanying source. The CDDL and GPL are fundamentally incompatible in such a way that it is quite difficult if not impossible to distribute binaries derived from mixed CDDL/GPL code without violating some clause of one or the other license. I won't claim to have an expert-level understanding of the details, but I work with several people who do.
.
Now, as for why "someone" doesn't distribute some magical patchset which will allow you to build a ZFS-enabled Linux kernel... It's because "someone" doesn't feel like keeping up with the massive amount of work required to maintain said patchset against ZFS and Linux kernel changes. The work would not get nearly the amount of exposure and testing necessary to make it a first-class filesystem option, and so would languish as a hobby/experiment with not much payout.
.
Much better to just focus on btrfs. Or wait and see what happens after the acquisition completes.

Let's ask the customers...

[Catalina588]

About five years ago, Sun seriously considered killing Solaris on X86. What Sun customers said was "Do that and your SunFire servers will be out on the street as quickly as we can get them unplugged". These customers included the major NY city investment and merchant banks, and Sun was in no position to destroy relationships with key customers. Fast forward. Those same large financial institutions are still running Solaris, including Solaris on Sun-supplied X86. But the reason this rumor makes no sense is that Sun and Oracle grew on Wall Street together as less expensive alternatives to Big Iron IBM mainframes.

Kill Sun Solaris and Oracle commits suicide. Makes no sense at all. Won't happen.

Re:Let's ask the customers...

[PHPfanboy]

Why destroy relationships with key customers, when your key customers can destroy themselves! Yay, go Bankers!

The Suns Connection Plan ~ FTHM

[Anonymous Coward]

I recenlty spoke to a sun sales rep at a BBQ
And he says that sun tied everything that they made to everyting else to its highly unlikely that it will go dead.
Most likely scalled back

Cut And Paste

[Anonymous Coward]

Wow... does Steven J. Vaughan-Nichols actually write stuff?

I thought he just cut and pasted from press releases.

See what happens

[bill_mcgonigle]

You're right, and Oracle could hasten this: re-license OpenSolaris under GPLv3 (patents) and see what happens. Worst case, nothing.

Likely case: the community ports everything great to Linux and they don't have to worry about what to do.

I have to say, my one Nexenta box is very impressive and Linux does have some work cut out for it. Other parts are, eh, somewhat annoying.

I do hope Sun's documentation team stays on - they do such a great job.

Always wondered about this

[trawg]

...ever since Nullsoft's WASTE was released and AOL (or whoever) pulled it and revoked the license. From http://www.nullsoft.com/free/waste/ :

If you downloaded or otherwise obtained a copy of the Software, you acquired no lawful rights to the Software and must destroy any and all copies of the Software, including by deleting it from your computer. Any license that you may believe you acquired with the Software is void, revoked and terminated.

It was released under a GPL license (IIRC). So they have effectively revoked the license. They haven't tried (actively) to stop redistribution - indeed, there's forks on Sourceforge. I think Asus or someone even made a derivative product from it?

Re:Always wondered about this

[jacquesm]

Helpful of them provide the filenames of the sources...

http://www.google.com/search?q=waste-source.tar.gz&ie=utf-8&oe=utf-8&aq=t&rls=com.ubuntu:en-US:unofficial&client=firefox-a

Re:Always wondered about this

[jonbryce]

Justin Frankel didn't have permission to release the source code under the GPL in the first place as it belonged to his employer, not to him.

Re:Always wondered about this

[Mprx]

That wasn't revocation of a license. WASTE was a work for hire, so the authors never held the copyright and never had the right to GPL it in the first place.

So in the title...

[thePowerOfGrayskull]

... in the title we start out with "Mass speculation suggests...". In other words, "A bunch of bloggers are working hard to convince each other that Oracle wants to kill OpenSolaris". Why is this news again?

Kudos to Oracle

[lems1]

Bravo!

I'd hate for the technologies behind opensolaris/solaris to go away, but I'm all for cutting the excess in efforts to achieve the same goal.

If we can only get all the opensolaris things (kernel stuff, some of the user-space stuff) neatly bundle into Linux... Kind of some weird breed of Nexenta+Linux(kernel)+Ubuntu (which Nexenta already does since it's mostly based on it).

I just HATE the way everything is so different on Solaris/OpenSolaris. And I LOVE some of the kernel specific things that Solaris brings (dtrace, zfs, ... take your pick).

Kudos for Oracle if they get this done and focus on other things that works better for them. Next stop: MySQL. Yes, this also needs to be on its own community (like Debian), not controlled by a single entity.

It's not the OS that will die, it's SPARC

[Anonymous Coward]

It's SPARC that's gonna die, not Solaris. The OS is open and can/will survive *IF* its users want it to.

However, SPARC is doomed. Now that PC's can address 384 GB or more, and cost a fraction of "big iron", SPARC has little to offer that cheaper platforms can't provide. Sun are already competing with their SPARC line with intel/amd hardware. Why should Oracle support the huge R&D of those platforms, when there are 10 times as many Intel-savvy engineers to design cheaper 64-bit intel/amd systems...?

Yeah, you might say "but then Sun would just be another box-shifter, with their own OS!" Yeah, and the implication is that Sun are dead. Bye-bye, I never liked you much anyway.

Re:It's not the OS that will die, it's SPARC

[Anonymous Coward]

Oracle runs best on Solaris, Solaris runs best on Sparc. Sparc scales much higher than x86 (vertical scalability, not horizontal (aka clustering) scalability). Oracle backs and uses Java heavily to. Java runs best on Solaris. Solaris runs best on Sparc. Do the math, it's quite simple.

In terms of sizeable enterprise deployments, Sparc will actually cost you quite a bit less than an equivalent x86 cluster, not only up front on the hardware, but also in terms of space, electricity, cooling, etc. Sparc only costs more when you comparise the cost of a single T2 CPU to the cost of a single x86 CPU, you're not getting anywhere near the kind of throughput on a single quad-Xeon (4 cores for 8 threads) as you are on a single T2 (8 cores for 64 threads). Compare the cost of a single T2 to the cost of the amount of Xeons it takes to match the performance, and the Sparc wins out on cost. In fact, the difference on the up-front cost of hardware buys your Oracle EE license, with interest.

Sun didn't get into the x86 market because Sparc sucks. They got into x86 because they wanted in on the low and middle end markets. Solaris was tailor made for Sparc, Sparc was tailor made for Solaris, both were tailor made for specific workloads.

Sparc isn't going anywhere.
Solaris isn't going anywhere.

Missreading

[Amiralul]

I've first read: "Mass Spectrometry Suggests Oracle May Kill OpenSolaris" and for a few milliseconds my bain tried to find a casual relation between mass spectrometry, Oracle and OpenSolaris.

Big thanks to the guy behind OpenSolaris.

[Anonymous Coward]

I have used OpenSolaris to investigate problems for our customers thats using Solaris.
The value of having access to the code when things blowup is huge.

At the same time, I want to thank Sun for releasing source for the JDK.

From reading the comments made in the code, you can see that the coders has had
a good time working on Solaris. Some parts are actually really fun to read.

However, some parts are not that fun to read. Cause when you find a label named
"whycantjohnnyexec" you start to wounder why they never fixed the problem.

"Is Oracle getting ready to kill OpenSolaris?" FUD

[Zubby]

Blog response by Ben Rockwood - Quite a good read if you want some facts in your stories. http://cuddletech.com/blog/pivot/entry.php?id=1047

Re:Already Open- SAPDB/MAxDB case study example

[emes]

SAP closed-sourced SAPDB/MaxDB.

They never provided anything resembling a decent explanation as to why.

kernel merge?

[shish]

Step 1) Relicence opensolaris in a linux-compatible way, then dump it in a tarball somewhere and forget about it for a while
Step 2) Wait for open source developers to port all the useful features to linux for you
Step 3) Profit!

Would it make more sense to kill Oracle Linux?

[walterbyrd]

I have had worked at four different companies that use Oracle, and they all ran Oracle on Solaris. I have never seen one installation of "Oracle Unbreakable Linux."

It would seem to me that Sun will be a much bigger part of the merged company than Linux. If something has to go, shouldn't it be Linux.

BTW: I personally like Linux better than Solaris. But I am trying to look at this from a business perspective.

(Linux is the biggest fish) = inspiration

[Anonymous Coward]

"The typical geek who builds a freeware application builds it for Linux first since Linux is the dominant freeware operating system."

I think it's all about statistics.
For me it's mainly because it's free and advertised by its fans as the only true free operating system with bells and whistles whereas other communities are smaller and not funded by IBM/HP/Red Hat/Novell and other *big* corporations.
I know many Linux users who keep criticizing every other UNIX-like systems and sometimes without having actually used them (They kept talking about "Slowwwwwwlaris).

I started with Linux because there was a Linux user group in my school and after 2 years I disliked the way GNU/Linux systems are developed and documented and switched.
I tried FreeBSD (+ other BSDs as well) and (Open)Solaris and found that almost every claims about slowness, unfriendlyness, lack of ergonomy were mostly *false* and only biaised claimed.
It's now +5 years since I discovered the UNIX way and I can draw some personal conclusions about what I like and dislike.

- Sometimes one is slower, sometimes one is faster, but what is the point of being faster when you must spend much time with configuration ?
- I dislike silent corruption of my filesystems with GNU/Linux.
- I dislike my system hanging because my RAM is corrupted and Linux does not check it whereas Solaris detects corrupted blocks and discard them.
- I like the documentation of *BSD and Solaris, the "logical" way things are organized, the ergonomy of command line utilities (pw under freebsd and zoneadm, svcadm, dladm)
- I dislike the way Linux distibution assert that "something works (tm)" when it's actually not the case !!!
- I dislike the lack of backward compatbility in Linux and the poor documentation of standards, and looooove well-written and stable interfaces.
- I dislike manpages in Linux and the facts that they sometimes points to websites (what is the point when you have no access to a network !), I like when they actually point to references and standards.
- I dislike the USB framework in Linux and the way devices are unconsistently supported from one minor version to another.
- I dislike the fact that my webcam is not recognized under Solaris.
- I like Dtrace and the "verbosity" of Solaris.
- I looooooooooove ZFS under Solaris and FreeBSD.
- I dislike the lack of simple how-to under Solaris.
- I like when a device is "not supported" by Solaris and actually works perfectly !!!
- Last time I upgraded my Debian systems it has been completly broken, last time I upgraded my OpenSolaris system
- I dislike ALSA (very much) and looooove OSS/Boomer.
- I dislike "lot a of features" without respect of standards and lack of documentation.
- I don't like when my GNU/Linux dies when swapping (no ssh)-

See... to me, every system has its plus and cons, and every community has its way of doing things.
Actually I like OpenSolaris the most because of modularity, consideration of standards, and higher level of abstraction (!!!) and think that "it gets better every day".

Now I think the learning curve seems lower with Linux because you can find many how-to's so that you can copy/paste (without sometimes understanding what you do). (call it "momentum")

People I met in Solaris and FreeBSD communities are all kind and smart as most users in the Linux community _id_ _est_ apart from some *intolerant* people who think they are the only enlightened ones.
More important Sun devs are friendly and eager to share their knowledge.

Hey all ... please do not consider all the FUD and Marketing garbage and use the system you like and enhance the "libre" eco-system and its bio-diversity !
I use OpenSolaris in the field of Scientific Computing and as Desktop (24/7), and don't feel like switching back. >_

Long live the OpenSolaris community and other communities :)

-- jollyd

lesson taken

[Anonymous Coward]

Considering the number of people proposing Solaris to be GPL'd so that ZFS and Dtrace can be ported to Linux ... an immediate conclusion could be drawn: Solaris features are valuable whatever some may say.

Then if Oracle decided to favor and support Solaris it could own a home-made solution (full sw stack) proposing several advantages over "corporate" Linux solutions (say RH,Suse) ...
I assume that its underlying cost would be negligible in regard of the advantage.

OpenSolaris is a great way to experiment before integrating in a "rock solid" version.

I don't see the point in discontinuing either Solaris or OpenSolaris

Reliance can make a license irrevocable

[mrmike37]

Having one outside developer work on an open source project might be enough for reliance to make a license irrevocable. I otherwise agree with you. I think it's very unlikely that the license is a contract because it would have no consideration (i.e. what is the user giving up?) Back to studying for the Bar for me!

Or just maybe ...

[dave87656]

Or just maybe ... it might just be in Oracle's best interest to provide an OS platform, which they control, totally optimized for their DB from the kernel on up. Add Sun's servers into the mix and you've got a high-performance h/w, s/w environment tuned for Oracle. They can't really accomplish this with Linux, but they could with Solaris.