Thou dost protesteth too much, methinks …

I read an amusing article linked to by the fine folks over at slashdot. In the Infoworld article that slashdot points to the title sets the tone. It is entitled “Linux will get buried”. I am going to look at this from an HPC viewpoint.

Apple is, and will remain for the forseeable future, a hardware company. All the software that it does, it does for no other reason than to sell hardware. OSX is a necessary “evil” for Apple, lest they really directly compete with Dell, IBM, HP, … . This is not to say OSX is evil, don’t read it like that. What is being articulated is that OSX is absolutely necessary for Apple to be able to sell its hardware, and carve out its small niche.
And I want to stress that it is a small niche. IDC is not seeing a troika of systems. It is seeing a rapidly growing pair, with one outpacing the other in terms of growth. And MacOSX presenting great growth numbers. Thats the thing about small markets, great growth numbers are quite possible/reasonable. MacOSX saw 16% CAGR to 4.8% overall market share for its line of machines. This is good. 5% of the market is nothing to sneeze at.
On the other hand, IDC’s predicted growth rate for Linux puts it at 7% in 2008. Even if Mac could sustain its 16% CAGR, it would still be behind Linux in terms of installed base in 2008.
If that weren’t bad enough news for the /. linked article’s thesis, another IDC quarterly study came out and is summarized here. You will notice that there is no Apple to be found in these numbers. Further some baseline data:

  • The fastest growing segment is the volume server segment growing at about 6.2% per year
  • quote: Linux server shipments grew 9.7% with the volume server segment representing the majority of both revenue and units.

There are a few other nuggets in there, and it is worth reading. The take home message is that a) small servers are growing fastest, and b) linux on the small servers appears to be growing faster than the other choices.
Ok, now the HPC context. There are as far as we know, a few MacOSX clusters. One or two large ones, several mid size, and multitudes of personal clusters. The big players, who are heavily invested in Linux and have their own hardware to sell, e.g. IBM, HP, Sun, Cray, Linux Networx, Penguin, Appro, … etc are not about to give up all their hardware revenue to Apple. It is at best, wishful thinking, to believe that the world of servers will be running MacOSX and windows, when the data show clearly shows the direction the world appears to be heading.
Remember, Apple is a hardware shop. They have not, nor will they likely ever open up OSX to run on anything but their hardware. Which is likely to be the reason why they have never broken out of their niche. And is the reason that they may never break out.
Dell makes lots of money selling servers preloaded with windows and linux. How much money can they make with OSX? Zero. And thats why it won’t dominate. There is no economic incentive for the large players to switch.
Of course, if Apple does something bright like make OSX runnable on non-Mac hardware, well, that may be a different story. I want to run my machines and clusters with an AMD64/x64/x86_64 ABI and set of chips. I can’t put OSX on my non-Mac laptop, or non-Mac servers. Which means that there is no way I can run OSX.
Don’t tell Microsoft this, but part of the reason for the success of Linux in the HPC arena is the element of choice, freedom, and open-ness. Vendor lockins are considered to be avoided if at all possible, at least, this is what our customers tell us. MacOSX is a hard vendor lockin, as is windows. The former is hardware and software, and the latter is software.
Then again there is the issue of commercial software availability as well. I don’t see the major players adding yet another incompatible platform when they have been working so hard to reduce the number to two.
All told, I think that article is an exercise in wishful thinking more than anything else. 5% is a good market share, and I would be surprised to see it more than 10% by the end of the decade. Unless Apple does something radical. Linux on the other hand, does appear to be gaining very rapidly on all fronts. Within HPC it already is dominant.

4 thoughts on “Thou dost protesteth too much, methinks …”

  1. I sort of agree on the hardware part about Apple. Strangely enough, I do think that OSX is their most appealing product. That said, I think that asking for OSX to expand beyond the desktop is probably asking for too much, and for that too happen, they have to go outside of Apple hardware (probably possible in theory with Intel chips coming into play).
    One problem I have always had with Apple, and why I will never own an Apple box, including an iPod is the vendor lockin. I am surprised this doesn’t get mentioned more often.

  2. Two points: when it comes down to it, OS is all about the apps. As far as I can tell, OS X will never have even close to the number of applications that Windows and Linux each have. It will never have major, major success–even on HPC.
    Second, I couldn’t agree more with mndoci about Apple and vendor lock in. In 2002, on a lark, I bought a PowerBook and an iPod (this was before it was clear it would be the dominant mp3 player). Since then, Apple has released 3 versions of its OS–and charged me for every one. They’ve changed the form factor of the iPod plug from industry-standard FireWire to their proprietary format–meaning I can no longer buy accessories for my iPod. And, of course, they’re selling music in a propietary format. Apple has a religious following, and like any religion, they do everything they can to keep their converts faithful (and tithing!).

  3. I have heard, though I can’t find/remember the source, that the vast majority of the Mac sales were replacements of installed base. I can believe that, and I can easily believe they get a few converts.
    I am not sure if I would call it a “religion” though. Mac users are vocal, and quite expressive. There is much to like in OSX. I have a Mac laptop loaner here from Apple to do some development on it. The problem I find is that while Linux and windows are really quite close in terms of usage mode for me, the Mac is quite a bit different and I have to wrap my head around it.
    But at the end of the day, you are spot on. Its the apps. The apps drive the rest of the choices, as they should. This is true in HPC and the rest of the market. I don’t see OSX being a large application market, either for shipping commercial apps or for HPC apps. The only customers that ask us about Macs are educational.
    That said, PPC970 was a great chip. It is a shame to see its main user abandon it. Then again the best solutions don’t always win. Sometimes they fall flat on their face or run up against an unbeatable marketing machine. I have seen this many times in the market, and some of our current product mix is due to marketing winning hearts and minds while technologically superior products failed.
    There is a great story by Arthur C. Clarke called “Superiority”. Should be required reading by marketeers.

  4. Er… that ACC story tells the story of how superior technology that can win battles winds up losing a war. Thats the relevance, not a random tangent.
    I see a rough equilibria in a few years between Windows and Linux. Not much else, though AIX/HPuX and Solaris may claim replacement revenue. An old customer who is running into some hard times now had a multics box up and running into the late 90s. Doesn’t mean there was a market for multics though some could spin it that way. It meant that there was some specific software they could not move off that machine. They were locked in, and they had a large business dependency upon it. I see the same thing for AIX/HPuX and Solaris. I don’t know much in the way of critical apps on MacOSX … so if there are (outside of press/graphics), please let me know.

Comments are closed.