Solaris v. Linux: The “I’m not dead yet” battle

The market has largely converged on two OSes going forward. Unix demand and sales have been giving way according to IDC and others for the past few years. Linux has been and continues to take market and mind share away from it. Most OEMs realize this. There was a legal battle over this, now preparing for the fat lady with the Viking hat to start belting out her tune.

And in a Monty Python-esque manner, one of the combatants says “I’m not dead yet, I think I will go for a walk”. Or maybe they are like the black knight at the bridge and say “tis but a scratch”.

Uh huh.

Sun, with their great record on Java working the same everywhere, as long as you don’t run on IA64, PPC, 64 bit anything, … wants to do the same for you with their OS. Thats right, they want to out-Linux Linux.

Linux, which runs everywhere (and I mean everywhere from cell phones to supercomputers) is being challenged by Solaris (well, OpenSolaris) which runs, well, on Sun approved ABIs/hardware. Lets let this sink in for a bit.

I couldn’t get a working driver for an extremely common NIC a year ago. And yet this is going to challenge Linux.

Its installation was horrible. Reminded me of old ancient Unix installations. Tuning it, configuring it was, again, hard. Installing software was, again, hard.

Ummm…. this is going to challenge Linux?

I have a question. Precisely how?

Linux has the driver mind share, the developer mind share (one platform to build for across a huge range of target delivery machines), the media mindshare.

Solaris is a throwback to the bad old days of a divided Unix. Theirs was “better” all the vendors would say. Yeah, and largely source level incompatible too.

Again, how, precisely, will Solaris challenge Linux? By offering more of the same?

ISVs, the people who write software, have to support it and sell it, have been clamoring to get off the zillion unix edition for a really long time. They are not thrilled with the prospect of Yet-Another-Platform. This is part of the reason why, despite press releases to the contrary, Microsoft is having so many problems getting ports to CCS. If the apps run on windows, why should the vendor port? It adds time and cost, with little benefit (none really, as CCS is not driving volume/large business). Solaris faces this issue and worse. It already runs on Linux, and customers are adopting this, in droves (see IDC for data).

Solaris is a legacy OS, still used in some areas, though, at every customer we work with and sell to, it is being cordoned off and phased out, in favor of Linux, with maybe one or two exceptions. In some part this is due to the costs of running it, in other part it is due to the high cost of software on Solaris. Absent Sun funding development efforts, the ISVs have no choice but to charge higher costs to the customer for the same product on Linux. We have just run into this in a number of reseller scenarios, where our cost from the ISV was 2x or more for the Solaris version of the same thing as the Linux version.

Again, precisely how will Solaris challenge Linux?

On drivers, Linux has won this, hands down. After installing my Dell laptop, and quite a few other machines now with both Windows XP and Linux, IMO, Linux wins the driver battle against windows XP as well. Solaris isn’t even visible in most cases.

On ISV market share, Linux has won this. Consolidation has closed off extra non/marginal-revenue ports (AIX, IRIX, HP-UX, Solaris) in favor of growing ports (Linux). This is what drives the market on the supply side.

This game was over a long time ago. The winner had been declared by the market. Unless an amazingly compelling case can be made for Solaris everywhere, from Cell phones through Supercomputers*, I expect that this decision will stand.

* Note: about a year ago I compared a highly optimized Solaris binary versus a moderately optimized Linux binary running on the same physical hardware, performing a HMMer calculation. Despite assurances that I was doing the right thing with the Solaris Studio compilers, the Linux binary was about 30% faster than the best effort Solaris binary. On the same exact source code, with the same exact input decks, on the same exact hardware. I dual booted the machine.

Viewed 9162 times by 2055 viewers

10 thoughts on “Solaris v. Linux: The “I’m not dead yet” battle

  1. Hi Joe,

    In regards to your observation that Linux binaries (not optimized ) run faster than Solaris binaries…..I totally agree. In fact this is a major reason why we (Platform) started supporting Linux – we had many customers running on Solaris boxes and they asked us ‘can we make your stuff go faster’ – we showed them the results of running on Linux and Solaris – they moved pretty fast to Linux after that.

    Now I’m not saying that Linux solves all problems but it is quite good.

  2. I realize that you’re making specific reference to Solaris in this posting, but let me say for the millionth time, “Not only is UNIX not dead, it’s numbers are growing.” Ever hear of an OS called MacOS X? 😉

    And keep in mind that “growth” does not equal “technical superiority.” I could list feature after feature of Solaris and AIX that Linux doesn’t have. Things like DTrace, the ability to lock a process in memory, etc., etc., etc.

    So pronounce UNIX dead and Linux the new king all you want. As long as there are people who want to use the better of the two operating systems alive, UNIX will never die…

  3. @Kevin

    MacOSX is a derivative of BSD ala Darwin. Linux is (not for copyright/IP reasons) a derivative of Unix.

    I have not seen MacOSX growth numbers *anywhere* outside of HPC, and there it is not growing at a noticable clip (from IDC data, bug them for details).

    I am personally not a fan of OSX, but it is IMO far better than most other Unix-y things out there, excluding Linux. Between the two is a matter of choice/preference. If I could run OSX on my Dell, I might try it. I won’t spend significant extra money for the privilege of running on Apple hardware.

    In an absolute sense, summing over all OSX, Linux, Unix (Solaris et al), yes it is growing. Looking only at Linux, IDC shows it is growing rapidly (fastest growing server OS). Looking at “Unix” (Solaris, AIX, …) shows it is shrinking quarter over quarter, at a fairly sustained rate.

    Sure, you can show me feature lists. Great. Wonderful. Simple question. Does the market assign a value to these features such that these features drive or continue to drive adoption? The answer to this question is a resounding “no”. So feature lists apart, we can pick out things in Linux that Solaris et al do not have (and won’t have) that customers are adopting. This cuts both ways.

    You are right, growth does not equal technical superiority. We can hold up Vista (and many other windows OSes there) as examples. Feature lists do not make something technologically superior. Moreover, superiority on a technical level tends to lead to people ignoring the market side of things. They think (at least in their minds and those who support them) that they have built a better mouse trap, so the world should beat a path to their door. This view is remarkably naive, in that it ignores that end users buy what works, is inexpensive, and so on. Some of the digerati *may* buy what they consider “superior” things, but largely the market is driven on price. “Superior” technology will often fall to an “inferior” one due almost entirely to marketing and distribution.

    Aside from this, “superior” is often a subjective analysis. I personally do not feel OSX is “superior” to Linux, though I acknowledge that others do. Regardless of whether or not in any measurable and meaningful manner, Solaris is “superior” to other OSes, is irrelevant as Solaris is being phased out of customer sites, due in large part to the cost of running and maintaining the “superior” technology. Ask Cray how much they like being “superior” some time. They have great stuff (IMO). Not everyone can get a desktop Cray. Their market is small and under encroachment from other systems.

    The only “Unix” that appears to be growing (and has been growing from IDC’s data, unless I am reading it wrong), is Linux and assuming you have data to back up the OSX bit, that. My understanding of the OSX market to date has been as a replacement for the previous generation of Mac’s. I would love to see contradictory data, so please, point it out if you have it.

    I am open to hearing about the OSX numbers, so please, if you have them post them. In considering where we need to port our tools, we ask which markets are large and growing rapidly. We will stay away from shrinking markets, or markets for which there will be little possibility of ROI.

  4. Well said! Very nice article. It’s an obvious thing slowlaris is a dead cow right now. Sun’s dead too. I suspect Oracle will stop supporting Solaris in the near future, because there’s no single advantage over Linux, but from the other side, Linux just kills it. Solaris is missing such important thing like RCU locking and it’s simply obsolete.

  5. @anon:

    Not sure I agree that Linux + KDE gives a “more powerful and innovative experience”. Apple is a master of the UI. I am not particular fond of some of their choices, but they do know what they are doing. I am also not fond of KDE’s user interface in general.

    These personal preferences aside, apart from ipad and iphone (and ipod …), it wouldn’t surprise me in the least if Linux had a higher real market share (e.g. actual desktop/server usage versus others) than OS/X.

    I am not sure one needs to “kill” OSX. OSX doesn’t quite suck as an OS … yeah, its limited to Apple hardware, and its not open. But it really isn’t bad.

    Yeah, my iphone is a great PDA (and terrible phone). Yeah, I’ve crashed it. No, really, iphones do crash. They aren’t perfect. But they are decent, and the apps are pretty nice.

    Android (Linux kernel + different userspace) looks quite good as well … but most of the phones I’ve looked at with it have a really crappy keyboard.

  6. @anon:

    “slowlaris” is somewhat negative. My main problems with it are an ancient user space, a terrible management/package management system, poor driver/hardware support, absolutely atrocious 1980s style installation system. Opensolaris made great strides in many of these aspects, getting to the point where it was possibly usable.

    Oracle may be simply keeping Solaris to itself for its hardware going forward. I dunno. The lawsuits from Netapp are definitely helping drive customers from considering it.

    Solaris had momentum in the mid/late 90s. Linux has it now. It might change yet again to something new. Who knows.

    No OS is perfect, no one will meet all needs everywhere. Though I will admit to being amused that hand held phones run similar kernels (same source for the most part) to the worlds fastest supercomputers. This is a testament to something.

  7. Regarding OSX, you have to keep in mind that it can be, umm, unpredictable in what you may expect from it:

    http://developer.apple.com/mac/library/documentation/Cocoa/Conceptual/CoreData/Articles/cdPersistentStores.html

    fsync in Mac OS X: Since in Mac OS X the fsync command does not make the guarantee that bytes are written, SQLite sends a F_FULLFSYNC request to the kernel to ensures that the bytes are actually written through to the drive platter. This causes the kernel to flush all buffers to the drives and causes the drives to flush their track caches. Without this, there is a significantly large window of time within which data will reside in volatile memory???and in the event of system failure you risk data corruption.

    A then MySQL developer (now doing Drizzle for Rackspace) has said that the above only applies to internal disks, it works differently for external drives on OSX..

Comments are closed.