NYT Ponders the Future of Solaris In a Linux/Windows World

JerkBoB links to a story at the New York Times about the future prospects of Sun's Solaris, excerpting: "Linux is enjoying growth, with a contingent of devotees too large to be called a cult following at this point. Solaris, meanwhile, has thrived as a longstanding, primary Unix platform geared to enterprises. But with Linux the object of all the buzz in the industry, can Sun's rival Solaris Unix OS hang on, or is it destined to be displaced by Linux altogether?"

Comments

Of course

[Anonymous Coward]

The current solaris systems will only have issue with this if they actually need to be rebooted one day and the new admins notice its not linux.

Re:Of course

[dnoyeb]

Thats a technical reason. In the US though, its about shareholder perception. If the CEO is trying to do big things and keeps talking in "old" terms, he will quickly be accused of being out of touch with reality.

Like the post said, Linux is generating the buzz. The money follows the buzz even if it does not make sense.

Nahh.. its not going anywhere.

[k1e0x]

Solaris is a great big iron OS. I don't think it will be disappearing anytime soon.

Virtualization makes Solaris less relevant

[compumike]

The purpose of the operating system is to act between the hardware, system abstractions, and the algorithms. But now that virtualization is taking over, the hardware responsibility of OSes is being minimized -- or centralized. Therefore, the advantages of one hardware platform can be more easily decoupled from those of an OS.

In my opinion, Sun was always known for rock-solid hardware, and this move toward hardware-agnostic computing means that Solaris gets just a bit less relevant today. Especially since cost is still a factor, and the hardware-specific advantages are disappearing...

--
Hey code monkey... learn electronics! Powerful microcontroller kits for the digital generation.

Re:Virtualization makes Solaris less relevant

[mysidia]

The purpose of the operating system is to act between the hardware, system abstractions, and the algorithms. But now that virtualization is taking over, the hardware responsibility of OSes is being minimized -- or centralized. Therefore, the advantages of one hardware platform can be more easily decoupled from those of an OS.

It's more important than ever, and Solaris delivers virtualization through Sun logical domains, hardware virtualization (zones), and the xVM hypervisor based on Xen (with Solaris dom0).

Now with Virtualization: if your physical machine fails, and your hypervisor cannot recover immediately from the error (without a reboot), you don't just have one server down -- you have many virtual machines disrupted.

Your Linux VMs that use ext3 may have some problems after that unclean death (filesystem corruption); whereas, your Solaris VMs backed by ZFS are less likely to have fatal problems, as filesystem backed by ZFS with redundancy 1 with no issues, as ZFS all but guarantees filesystem itself is consistent.

You can restart axed VMs on another server, perhaps: you have a shared storage environment, but this disruption costs something, and it is higher the more VMs you were running on that machine.

Also: there are application I/O-intensive workloads that do not virtualize well, such as high-load databases i.e. your 5 billion row Oracle DB.

Solaris is perfect for managing the hardware for these specialized applications that are not efficient to be run in a virtual environment.

ZFS may also be a major factor.

If you have shared filesystems -- you need SANs & NASes.

Do you use a $5 million NAS, or do you buy a couple high-end Sun servers, load them up with a direct-attach storage arrays, and share via NFS in order to provide a good bit of reasonably fast storage on the cheap? hmmm...

Think about it: Linux isn't really even a possibility for backing your servers' storage, you want 99.9999% uptime of your servers, right?

Systems based on Linux cannot generally guarantee uptimes like that (yet); although it is over time getting closer.

Your storage can't be efficiently shared by a VM (I/O in a virtual machine is notoriously slow, plus the CPU usage and available memory for caching is limited by other usage of the host).

Re:Virtualization makes Solaris less relevant

[C0vardeAn0nim0]

just atended a seminar about solaris LDOMs at the office. with LDOMs (which only run on niagara T1 and T2 CPUS) you can assign a whole PCI bus to a single guest OS in such way that the guest has full control of that bus, which implies no performance penalty if that bus have a couple of HBAs to connect it to a storage/SAN.

the sun guy at the seminar (same instructor that ran the solaris 10 administration course I took last year) made it very clear that the kind of hardware backed virtualization provided by LDOMs is the second best for high I/O apps, losing only for hardware partitioning, and it's way ahead of the kind of full software virtualization provided by vmware, virtual pc or virtual box.

Re:Virtualization makes Solaris less relevant

[phantomfive]

That is the problem with Sun: they make the most amazing systems, for a market that doesn't exist. They have systems that can hot plug processors. If that is not I cool, I don't know what is. But then if you have a cluster of $200 machines, and you can just replace the whole machine, then the usefulness of such technology gets less and less, unfortunately.

solaris is the new AIX

[timmarhy]

it'll still be around because it's a good OS, worthy of being used.

it'll just be a niche product.

personally i think it's sad sun blew their chances with solaris, it's superior to linux in security and performace.

Solaris 10

[Tsunayoshi]

If your only experience with Solaris is v8 or v9, you really need to check out Solaris 10. It is a complete night and day difference in ease of use and features. Add to that the volume of useful enterprise management software from Sun (the N1 stack, and now the new xVM stack) and you have an enterprise that is a dream to maintain.

I've been doing straight Solaris 10 admin for the last 2 years (linux for 4 years before that), and shortly will once again be taking a position that will be 99% linux. I will miss Solaris 10. I still love both OS's, but Solaris wins in my book at the moment.

Re:Solaris 10

[segfaultcoredump]

I agree.

If we were talking Solaris 8 or 9 vs linux, it would be no contest. Solaris 10 is another story.

In the current shop where I work, Solaris is the OS of first choice. We only run linux if we absolutely have to due to application compatibility issues. From a cost standpoint, they both run on the same server hardware (intel or amd), so there is no cost advantage there. The major one is that a RedHat support contract cost more than the equivalent one from Sun for Solaris (they are both free if you dont want support... sorta.... You can download solaris from sun for free and install it. You cant do the same with RedHat. That said, if you use kickstart you can get around the mandatory rhn registration that stops you dead if you dont have a valid support contract.)

  It just that with solaris we have the ability to load the box up with dozens of zones that are easy to manage compared to the alternatives on linux (yes, there are alternatives to solaris zones, and all of them involve unsupported kernel patches. ) We tend to run 20+ zones per dual proc box. Each zone gets its own env. We dont need them to differ too much from the main zone, so things like Xen are overkill. Chroot would be nice if we could also get better control over things like IP's, admin access, hostname resolution, etc. For what we want, Zones are absolutely perfect. zfs and Dtrace just add icing to the cake.

That said, this article read like a BMW sales rep's opinion of the newest Audi. If I want opinions of if solaris is dying, I'm not going to go and ask the head of the linux consortium which has a vested interestin seeing their prophecy come true.

  For a good time, go to the IDC site and read the comments. Most of them are ripping the author for the piss poor job he did since it reads like a linux marketing piece more than an actual news article.

Probably the wrong question.

[fuzzyfuzzyfungus]

I'm not sure that asking "will solaris survive?" is the right question. Any server OS with decent legacy traction can hang on for ages even without exciting benefits, or even parity, compared to its competitors. Any OS can also be opened up, given away, and allowed to limp along for as long as anybody cares to play with it. Solaris is essentially certain not to die.

The real question is "how much of a premium will Solaris be able to command?" This is probably connected to the question of how much of a premium SPARC hardware can command. If Sun gives Solaris away, and doesn't charge more than any of the major linux vendors for support, then Solaris will do fine; but that isn't necessarily helpful to Sun. If Solaris can justify a premium(either upfront or for support) or can drive or be driven by purchase of fancy SPARC boxes, then the resulting market share may be about the same; but far more valuable. That seems like the more relevant question.

How about some technical analysis

[synthespian]

Tech shouldn't be about "gee, everybody's using it."
How about some hard, technical facts?
So many things in Solaris are more advanced than Linux...Sounds like a Linux PR piece...
For instance, you can count on general ABI breakage on Linux. They even take pride in it. That's not a system you can trust for the long haul. You can't trust your applications will remain compatible.
Linux is a mess, IMHO.

Re:How about some technical analysis

[Darkness404]

Tech shouldn't be about "gee, everybody's using it."

Ok, lets switch our servers to Plan 9, surely any CS grad will understand how to perfectly use it, right? While yes, you are correct in saying you should look at the benefits of Solaris Vs Linux, and because it is still UNIX it is more or less a void point, but if most of the new graduates don't work with an OS, it is bound to die out and get replaced with a more familiar OS. Either that or you have an incompetent admin running your servers.

That's not a system you can trust for the long haul. You can't trust your applications will remain compatible.

No, but you can usually count on a more up-to-date system and generally a stable one at that. Solaris only has a few releases, Linux has many, many more (mostly because it is OSS). Also, most UNIX apps are developed for Linux and later ported to Solaris, not the other way around, meaning that some things may be untested on the Solaris platform.

Not all the best features are technical

[mcrbids]

When evaluating the success of a product in the marketplace, it's important to note that there are many features of even highly technical products that are not technical in nature, at all.

Linux compares very closely to BSD from a technical standpoint, BSD has a much longer history than Linux, and is arguably better than Linux in many areas. It's definitely had more time to mature. So what feature does Linux have that has everybody talking about Linux?

Its license.

I'm not knocking the excellence that Linus Torvalds has displayed over and over again over the years. He's done a great job and I depend on his efforts every day in running my own business. But as great as Linus has done managing the technology of Linux, it would be hard to say that Theo De Raadt has done any worse. It would be easy to claim that Theo's work is more secure, but both have produced excellent products that are truly world class in nature.

But what has everybody talking about Linux is the license - the share and share alike requirement laid down by the GPL, which turns the Tragedy of the Commons around on its ear so that everybody is pushing the project along together, rather than taking what's convenient and giving nothing back.

The sad truth? "More free" isn't always better. Just like "less government regulation" isn't always a good idea, you can often get a better mix for everybody by limiting people's freedom to screw each other.

Now, Solaris is behind the 8-ball. Even with the same license as Linux, they'd have to show a clear, compelling advantage to cause people to switch their efforts away from Linux. Given just how good Linux is in so many different areas that Solaris can't even touch today, that would be very, very hard to do.

Show me a Solaris supercomputer and I'll show you hundreds of Linux-based supercomputers. Show me a $40 Solaris-based router, or a Solaris phone, or a Solaris-based pocket calculator. Ironically, while Solaris is touted for "big iron", it's a non-starter in the list of the top 500 supercomputers, while Linux is dominant.

Go Tux!

Re:Not all the best features are technical

[the eric conspiracy]

Solaris doesn't play in the supercomputer market because the Sparc architecture is not cost effective for that application. The big iron that Solaris runs on are enterprise scale database servers, which are optimized for an entirely different set of performance parameters.

Re:Not all the best features are technical

[BlueQuark]

Please elaborate with what aeras and tasks that Linux is so good at that Solaris
can't touch?

I use both Solaris and Linux in my environments and Solaris 10 is by far the superior
OS in my opinion. We have Solaris servers on both SPARC and AMD64 and Linux on AMD and Intel 64 bit hardware.

We had migrated a number of Sybase instances to Linux, but we kept having reliability and performance problems, so we migrated them back to Solaris but Solaris 10 on AMD64 boxes
and we've been extremely happy with the results.

Our company is current migrating all of our market data servers to Solaris AMD64 servers in Zones and will reduce the number of Linux servers which stand at 25 to 4 X4600s running Solaris 10. In our testing of Solaris on x4600 as opposed to DL585s (same CPU and memory configurations) we have seen a large performance gain and cheaper operating costs, since we don't have to pony up a RedHat license for each server.

Re:Not all the best features are technical

[Waffle Iron]

It's not the license. It's the huge PR machine behind it. IBM and all big boys supporting Linux are really selling hardware and are using Linux to commoditize enteprise software.

And do you know why these companies chose Linux as the focus of their PR machines? As the GP post argued, it's because the advantages the GPL provides to these big companies over BSD-style licenses.

These big companies know that if their competitors use the code they contribute, it will have to be on the same terms as the contributor. It can't be closed up, embraced, extended and then used as a weapon against the original contributor. It's a simple exercise in game theory.

Rival what huh?

[Aphoxema]

I wasn't aware of the 'rivalry' between Sun and, uhh, those bunch of other people who openly contribute to GNU/Linux.

Maybe it's similar to that 'rivalry' between Gnome and KDE, or Slackware and Red Hat, or all those other things that it's generally the onlookers that assume there's a conflict because heaven forbid there's such a thing as two different things sharing the same space when there's a choice to be had between them.

So, one day Solaris might win and then everything else will be gone?

These competitions only exist when there's money or ratings at stake, or when people are bored.

Of course it isn't going anywhere

[debatem1]

Solaris runs on mission critical systems, the kind that will (absent horrible, business-destroying failure) be running until their tape drives rust solid. It isn't going anywhere, and even if Sun's interest in it evaporates, there will probably still be thousands of systems out there chugging along merrily for years to come.

Re:Unification

[Darkness404]

One or at most 2 will become defacto standards and the others will fade away.

We already have that. We have Red Hat (RPM) based distros and Debian (APT) based distros. Just about any major distro can fall into that category, CentOS, RHEL, Fedora, Yellow Dog, and even openSUSE can be considered to be with Red Hat based (RPM) distros. On the other hand, Ubuntu, Debian, KNOPPIX, Xandros, DSL, etc.

There has also been a lot of talk about LSB that could help unification (which, honestly IMO is not needed and will just be a waste of work on distros for a failed standard)

MacOS could become dominant

I can't see Apple wanting OS X to become dominant. They make $$$ of of hardware sales to fanboys. The die-hard Mac fans. Apple honestly wouldn't be able to keep up with the demand if Macs had 25% or more of the marketshare. Apple is happy to sell iPods to everyone and keep the Macs for the fanboys. Now, they want OS X to have access to all major software and to have drivers, so they don't want too low of a marketshare, but I can't see Apple wanting OS X to have more than 10% of the marketshare. Much as how Ferrari doesn't want us all to be driving Ferraris, it loses the prestige of driving one.

In the worst case scenario the desktop and server segments become so fragmented that you'll have dozens of versions of each app - 1 per OS.

Ummm... How is that bad? There are dozens of versions of Apache, one for each OS, yet it still manages to be a unified server. And dozens of separate distro specific Linux kernels but just about all are compatible with all programs (when the proper libraries are installed).

Old OSes don't die, they just fade away...

[TheMidnight]

Solaris is the smallest percentage of UNIX platforms my company's clients run on. AIX is first, followed by HP-UX. However, though Linux is a popular operating system with universities, web sites, startups and small server solutions, Linux on x86 scales horribly (and I do mean horribly) on our application and other high-performance database solutions with thousands of users compared with the big UNIX operating systems. ext3 can't support the filesystem throughput required even with RAID 10.

We still configure Solaris systems on Solaris 10 UltraSparc, and I believe Sun just came out with a new, rather mean processor. Solaris, and certainly HP-UX and AIX, are not going anywhere soon. There are too many enterprise database systems (new, not just legacy) that require the far more powerful and scalable hardware and software that Sun, IBM and HP offer.

Have you ever benchmarked the 4.7 GHz POWER6 chips on AIX 6.1? It's the fastest processor and operating system combination I've ever seen.

Re:Performance

[Anonymous Coward]

Got the balls to drop an strace on your production Oracle database? I tried strace on an Oracle database on RHEL 5 and the damn process deadlocked and the box needed a reboot to clear it up. Good thing it was a development DB.

I've put a truss (and now dtrace) on PRODUCTION Oracle databases running on Solaris many times.

I don't dare do that on Linux.

Solaris is as far beyond Linux in stability as Linux is beyond Win2K.

Solaris and perception

[Anonymous Coward]

The real question is "how much of a premium will Solaris be able to command?" This is probably connected to the question of how much of a premium SPARC hardware can command.

Sun sells some (really nice) x86 kit. Solaris is certified and supported on HP hardware (though HP is not an official OEM). Dell has an OEM agreement with Sun, and so does IBM. Furthermore Solaris is being ported to IBM's mainframe systems, and it works just fine as a guest in VMware (and xVM, and work is being done with Xen).

A software support contract is cheaper for Solaris than it is for Red Hat.

The main issue is perception: Solaris is viewed as "old and tired", and Linux is viewed as new and exciting. I do not think this corresponds to any meaningful reality (and I've run DOS, DESQview, OS/2, Linux, BSD, Solaris, and OS X on my home machine since I began computing).

My perfect system would be the core of Solaris, the interface of OS X, and FreeBSD's ports tree. The development model of Linux (and BSD and GNU/FSF), and the freedom it gives you, is the most important thing that Linux has brought to the table, but I don't see anything inherent in the technology that Linux gives that makes it anything special.

Re:Or else...

[countSudoku()]

I use the Solaris 10 JDS everyday at work and also run Ubuntu and Solaris 10 x86 at home, with zones on it. Basically the "apt-get" you're looking for is called "pkg-get" and is available from blastwave.org.
        The future of Solaris on the desktop is not as exciting as that of Ubuntu, or any other wildly popular Linux distro. The enterprise future of Solaris is way more exciting IMO. The reason is this; Solaris 10 Zones are ready for primetime enterprise whereas Linux is still being pondered and in most cases not being taken seriously due to the open source nature of the beast and the sheer number of different distros, many of which lack and enterprise level support. We had a Red Hat box in our DC and we retired it. Meanwhile we're approaching 400+ Solaris 10 zones and we're coming in at *HALF* the price of the VMware solution that the Windows side of the house provides. Guess who's growing faster? Even with more expensive boxen our solution is a better value and provides a very solid framework for many, many environments.
        As an OSS advocate and 20+ year SunOS/Solaris admin I will welcome both operating systems in my home and data center. Like the man said above, you know one, you practically know the other. It's all *nix in the end.

Re:Or else...

[udippel]

That was the main reason why I canned OpenSolaris after trying it out.

Uhh. Yes. No.
I guess you talk about the miserable 2008.5 or so.
I for one hopped on the SXDE/SXCE/Nevada; just for a try, and have been there ever since. It contains (almost) all that I need, out of the box, from Macromedia Flash to StarOffice. The most complete one I ever tried. Compiz included, and the (alas) proprietary blobs (NVIDIA). Show me a Linux that you can install, and brings up 173.14.09 NVIDIA on reboot, at the correct resolution?

Actually, it only missed on Mplayer, Dia and Gnumeric for me. Over.
Now I run a few multi-boot systems, with Debian/Ubuntu and OpenSolaris as choices, and I tend to use OpenSolaris. The problem of OpenSolaris is not so much OpenSolaris as it is the SUN management's low visionary angle.

True, hardware support is better on Linux(Ubuntu), resource requirements are easily 2 GB (it consumes close to 1 GB after boot and starting GNOME), networking, and especially wireless, is far behind Ubuntu. Mounting of ext3 needs external programs to be installed the old SUN-way, and is not possible as R/W; it is limited to readonly. For the rest OpenSolaris is at par, if not superior.

That default username "Jack" does not exist in the world of Nevada. As I said above, the one that you had your hands on was 2008.5; codenamed Indiana, enforced on the project by a bunch of managers at SUN who have since long passed their expiration dates.

It is not only the DTrace and Zones that make it great(er), ZFS is truly the file-system contender for the next millennium. A file system needs to be seen as a self-contained, holistic, entity. fdisk and format are so last millennium.

The Caiman installer was the greatest piece of installation software I ever used. But it was broken half a year ago and now rumbles on as a low-priority project. Don't talk to me about bullshit: The installer is the first thing a potential user hits. And currently it is either the venerable CDE-SUN installer or console. This is how to screw up a visionary project like OpenSolaris at its best.
And then, community. Make me puke. The bug reports that I file (and still file) disappear in the intestines of SUN, and I can't even see them. Not to mention modify them.

Yes, OpenSolaris might fail, in the end. But not on technical reasons, and this is where the article is erroneous. It is a badly managed hybrid of proprietary-free software.

Re:Or else...

[Joebert]

I have to ask, are you an email spammer ?

Re:Or else...

[Saint Aardvark]

Blastwave...heh. Which Blastwave are you talking about?

Sorry, this is a bit of a sore point for me. At work, we have a Solaris 10 machine that powers about 30 SunRays for mathematicians. JDS is fine, but adding other programs is a pain. (Disclaimer coming up, so bear with me.)

• Blastwave: They just had the split. But before that there were problems. Upgrading CUPS broke printing; they'd moved around some Ghostscript filters. Upgrading Postfix broke Postfix, because they'd moved the config files to play nicer with zones, and their script that should've dropped everything in the right place didn't. These were stable versions, not the unstable.
• pkg-src: Great until you trip over something that won't compile and spend days trying to track down what it is -- say, 1 package in 20. Sounds like good odds? Try compiling Firefox or Kile, with dependencies stretching back to libc and the Dead Sea scrolls. I'm guessing they just aren't able to do as much testing on Solaris...and fair enough; the job of making umpty thousand packages compile on mumble different OS' is hard enough.
• compile from source: fine, unless it's obscure (say, some mathematical package) that assumes GNU tools all the way, or a Linux OS, and weird, obscure things break.
• download binaries: yes, if they've got 'em.

And now for the disclaimers: No, this isn't enterprise (which was your point; I was looking for a place to jump into this discussion, and the mention of Blastwave got me). Yes, a real sysadmin could compile all this from scratch without problem. Yes, this is an edge case on top of an edge case (desktops for mathematicians? How obscure!). Yes, ZFS and dtrace are seriously, jaw-droppingly awesome.

But this is my experience; so far, I simply have not done anything remotely enterprise. It's all been server + desktop in small shops. And for that environment, requirements are changing all the time. The mail server now needs to do spam filtering and DNS. Yes, they should be split up, but there isn't the budget. The new guy wants KDE on his machine instead of Gnome, or needs to try out a new library to see if it works.

And for these, it's not "set it and forget it"; we need new packages, or updates to the old ones, all the time. If all the heartache I described was a one-time thing, I'd do it and be done...but in this environment, it'll need to be done again in three months. That means a good package manager (hello, Debian!), or a good ports tree (*BSD), or an environment that everyone is familiar with (Linux, because it has just that much mindshare).

Bit of a rant, and less coherent than I'd like. But it's 6am, I haven't had my coffee yet, and my kid's about to wake up...so I'll have to leave it there.