Slashdot Mirror
Linux Server Sales to Reach $9.1 Billion by 2008
dunric writes "ZDNet is reporting that sales of servers using Linux will reach a whopping $9.1 billion by 2008. Annual revenue for Linux servers is expected to grow by a healthy 22.8 percent, compared to just 3.8 percent for the overall server market. Additionally, Linux servers will account for nearly 26% of all server shipments."
of a free program. From this we can conclude unit sales of Linux to be about Infinity Times 9.1. That's pretty good.
As can be confirmed by simply going to Netcraft, Slashdot actually runs on the crushed hopes, dreams and spirits of thousands of self-proclaimed, social-anxiety-disorder-afflicted "nerds".
Methinks ZDnet published this prediction simply to exploit the predictable slashdot-effect response to such a story. I am projecting a 22% increase in ad revenues from their banner ads featured with this story at the tail end of 2004.
$5 / month hosted VPS on linux = awesome!
This news is interesting but what would be interesting to see is which markets Linux servers are eating up. People of /. are going to assume "FINALLY THE DOWN FALL OF MICROSOFT!". But the truth of the matter is Linux is probably eating up the markets of AIX, HP-UX, SunOS, Alpha, etc. Look at Sun's stock.... Down over 30% from last year? Even if this statistic is true I don't think Microsoft is probably losing any business but rather our fellow UNIX brothers. Go Linux!?
Keep in mind that the $9.1bn number is in hardware sales, not software.
The last eight Intel servers I installed were all assembled from good quality motherboards, fans and better quality ATX power supplies into run of the mill whitebox full/mid tower cases.
If space is not an issue then I find that taking time to assemble a well laid out PC case delivers better reliability than Intel based 1U or 2U rackmounted servers.
"A whooping $9.1 billion by 2008", or so it is reported.
But what about the Windoze servers ?
"A whimper $18.2 billion by 2008" ?
Muchas Gracias, Señor Edward Snowden !
Seriously, saying "Linux Server" to a CEO has one of two effects. One, they glaze over and continue using their familiar "Windows Server". Two, they think you're really smart and give you lots of money. Sure, there's the rare third case where they'll realize "Linux = OS, Server = Hardware", but chances are they're the CIO.
Does this mean I should open a "Servers The Run Linux" eBusiness? Amazonux.com, perhaps?
UTF-8: There and Back Again
Everybody knows that people buy Linux servers just so they can install pirated versions of Windows on them!
Linux may be sitting high and pretty on the desktop market, but it has to create a usable UI to break on thru to the server market.
That's what the adverts on Slashdot say, anyway
Slashdot: News for Nerds, Stuff that matters only to them
Whenever I see words like forecast and prediction buried, I wonder what the motivation of the writer is. I dug around a very little bit and found this link to an IDC ress release that this is based on (I think). IDC - Press Release http://www.idc.com/getdoc.jsp?containerId=pr2004_1 1_02_093312
Reading the press release, I could only think to myself "uhm, duh, who couldn't see that coming."
I do wonder how they came up with the time frame though. Four years seems like a long time to predict anything in the IT business with any accuracy.
the dollar amont is not as important as the number of units. 9.1 billion is like, what, 100 sun servers? seriously though, the numbers of servers shipped is more important. because alot of that will be replacing NT servers. and alot of that will be new server infrastructure. every linux server sold is one less windows server, regardless whether it replaces a sun/ibm or not. dollar sales are a relative figure. what matters if the total number of servers, or market share. and what matters is what they're used for. are they just serving up web pages or are they running the backbone of business web applications? if linux is relegated to the periphery, it won't matter a whole lot.
My problem? I was perfectly gruntled, until some numbnuts came by and dissed me.
This raises an interesting point. If pricepoint is genuinely what is the deciding factor for these predictions, what will happen when OpenSolaris is release?
From the article - This is not a troll, but I have never understood the wide-spread embracing of Linux to be a direct result of anything but price-point and community support. I hate near everything about the way Linux is structured when compared to other flavors of UNIX, and I am not a fan of the kernel internals, yet I keep going back to it because of the aforementioned reasons. To be fair, I was introduced to UNIX with SunOS and the BSD family before being introduced to Linux, and lately I've been sharpening my skills with AIX and some of their enterprise solutions, so I may be totally missing the point of Linux.
Let's talk hypothetical here - Let's say Sun releases Solaris under a nice license that satisfies everybody - the BSD nuts, the Stallman-worshipers, and the corporate players [bear with me here; I realize I'm treading fairytale water], and let's say the community loves it and starts hacking away at it like a hillbilly with a hatchet. Right there, the Linux pricepoint and community support is matched.
I predict in the future we will see some more UNIX versions opened up, specifically, AIX. This is based entirely on speculation and the late-night readings of IBM papers, but I wholeheartedly believe in the next 10 years, IBM will either completely open the source or share a great portion of it (barring a SCO victory).
I myself have always preferred commercial UNIX to community efforts (although the *BSDs are near and dear to my heart) and have used Linux out of necessity, not out of direct superiority to commercial UNIX. My point is that if (or when) commercial versions of UNIX (such as Solaris and AIX) match the benefits of Linux, Linux may be the kid without a gimmick. But again, this is based entirely on the premise that Linux's gimmick is limited to the two previously mentioned, so if I am totally missing something, would some more-informed Linux guru clarify
Eh, it's late. Too much RPG IV.
"You and your third dimension."
Who will be able to predict the market in 2008? With spam, viruses and hacker attacks escalating, and Longhorn due to be released... who really knows what the market will be like then?
I hereby place the above post in the public domain.
This is sales of Linux *SERVERS*, not the OS itself.
Good point as we all know these servers now run pirated copies of the superior Microsoft(C) Windows(TM) Server 2003(C)(TM)(R).
PS: Don't forget to click the banner on top of
Bill.
I don't need a signature.
ZDNET is just another publishing company that generally have b grade IT writers. Just look at the spam software show down. Hangon.. where is spamassassin.. maybe they didnt pay zdnet enough money to advertise thier product so it wasnt included.
Got a question about UNIX ask it here : Unix/xBSD Forum
Don't forget that FreeBSD is stealing in on MS and the other UNIXes as well.
Here, we're winding down Solaris and replacing it with FreeBSD.
(although patch management on BSD is an absolute PITA... portupgrade my ass. Give me Debian anyday)
READY.
PRINT ""+-0
Get it straight! It's lower TC0 - Total Cost of 0wnership.
Around 2000, Linux was already reported at over 30% and to rise even more.
How is it possible that it is expected to account for only 26% of shipments in 2008?
Easy: IDC changed their counting methods in the meantime, while the earlier numbers are about shipments, the current numbers are about revenue and only for server-hardware that actually ships with Linux.
That is correct:
So to make a long story short, most Linux server installations do not exist for IDC.
Isn't it funny that Windows always looks good in heavily distorted studies (TCO "studies", market share studies, etc.) while they no longer look so good in not-so easily distorted studies (like Netcraft)?
Of course IDC is quite smart, they talk about "sales" and they know that people will think about shipments/units and not revenue.
While the older numbers had some touch with reality, the current numbers are just nonsense. In reality Linux already accounts for a lot more than IDC wants us to believe.
I have seen Linux displacing other O/S'es for consolidation purposes, and usually Linux itself is not the driving factor.
For example, VMWare/ESX is gaining a lot of market consolidating hundreds of Windows servers (usually test/dev) down to a dozen or so Intel servers running VMWare/ESX, which is Redhat Linux running VMWare. But the Linux side of it is almost invisible. I have spoken to VMWare administrators who refused to believe that it was running on Linux.
Also, I've seen large Oracle databases moved from Sun or HP hardware, to IBM Intel servers running Oracle RAC on Redhat. In that case, Oracle and Intel platform are the driving factors. Again, the Linux is packaged sepecfically to run Oracle RAC, so server administration is minimal.
My point is that, in my somewhat limited experience, people are not purposely moving to Linux, it's more that vendors are packaging their products that way, and it makes sense in many cases.
Serving Suggestion: Defrost
The parent post is maybe not so far off the mark, though i'd like to posit a slightly different way of looking at things. I believe the situation would be clearer (for once!) if we wrote GNU/Linux instead of Linux.
Linux is just a kernel. Another child post mentioned it has good hardware support compared to solaris, i'm sure there are some other good points (e.g. a lot of architectures supported, embedded apps, a formidable base of experienced open-source developers, etc.,) that could mean Linux wouldn't just die if Solaris became free/Free.
But what most people think of as Linux, and what is in fact the largest part of, GNU/Linux is the set of userland tools we use. From the basics like bash, tar, grep, sed, awk, etc., to the compilers (gcc, etc.,), and up to the desktop level tools (KDE-family, Gnome-family, mozilla, openoffice, and so on).
Many of these tools can/are(!) ALSO be used on Solaris systems as appropriate/preferred.
If the Solaris licence is as free as the parent post hypothesizes, then this future is great! We can have a GNU/Solaris system if we want, Debian could offer a Debian-Solaris option (in the same vein as the Debian-NetBSD port), we can use bits of Solaris to improve Linux... All grist to the Free-Software mill.
Where I work, we all have servers as desktop development machines (typing this on a dual xeon, etc). We get our machines from Dell, and usually default to the Windows XP setup. As soon as we get them we reformat and install RedHat Enterprise or Fedora. I guess that means we arent counted in the stats, but instead are counted in the 'Windows Server' statistics, even though we are all using Linux servers. So I would think the Linux server stats are probably higher than stated.
Businesses tend to be risk adverse, which is generally a good thing. This means also that they are afraid of change. So this slows down Linux quite a bit.
Home users tend to stick with what they use at work. So until Linux takes over on the corporate workstation, it will be a slow tough fight.
All that being said, I think that Linux will kill windows. It will just be a slow process until a certain market share is reached. At this point application compatibility will be less of an issue. But progress is occuring much faster than some people realize: Linux is certainly killing proprietary UNIX (as is Windows), and the fate of OS X is uncertain, though I suspect that it will slowly be open sourced bit by bit, and they may slowly subsume eachother.....
Consider that 5% of the PC's which shipped last year ran Linux (mostly Linspire and Mandrake). Even after you count those where Windows was later installed, that was still up to three percent of *new* PC sales. Yes, Microsoft's monopoly has begun to collapse already. This year, maybe, it will be more.
Linux is already causing Microsoft real headaches in a few very key markets such as internet server and embedded system markets. The real beacheads are business web application development, desktop, and groupware now. But it is a slow process at the moment and will be for some time. I do predict though that it will be a fierce war for the desktop by the time Longhorn ships.
Also, Microsoft's last year of record profits was the year XP was launched. This is to be expected. But their market share is another question-- how do you measure market share? In dollars? If so then the slow demise of proprietary UNIX and Netware gives Microsoft greatly inflated numbers. If in deployments, then the simple answer is: we don't really know what real numbers are because we have no good way of measuring them.
Now, is there a tipping point? You bet. At a certain point, people won't write their business web tools for IE only (as Safeco does). Vertically targetted tools will be available for Linux, etc. and all basic productivity tools will be open source. At this point, I expect Linux useage to take off much faster.
ZDNet is reporting that sales of servers using Linux will reach a whopping $9.1 billion by 2008...
Considering Linux (and OSS in general) makes money via support offerings, shouldnt this be added to the overall $$$ amount ? Does 9.1. billion include support charges or simply the cost of hardware ?
Another thing I routinely keep hearing about is that hardware is going to keep become VERY cheap (as a matter of fact there were some articles suggesting it might even become free in the long term). If one cant sell hardware, and cant sell the OS, where the hell does 9.1 billion come from ? "Voluntary donation" ??
While that may be true, it has nothing to do with Linux - gnu tools worked fine on Solaris and BSD (and many other os's) before Linus was in secondary school.
Sent from my ASR33 using ASCII
Corporate purchasing has as much to do with IT directors furthering their careers by riding the trends than pure economics. With constantly increasing bangs per buck you can always do what you did 3 years ago for a fraction of the cost, but no-one gets that corporate promotion by doing the same thing as the last IT manager did. You've got to rip out UNIX and replace it with NT, showing the millions you saved ('96 to '00) or rip out proprietry UNIX and replace it with Linux ('01-'05?). People may see through the ripping out what you've got, replacing it with the new version of what you've got and claiming to save millions. Just leaving what you've got there and replacing it with newer, cheaper stuff when needed is for nerds with no ambition.
I think this article misses the fact that not only are people in the corporate world finally getting the issues with Linux in the enterprise, but by 2006 there should be an established new new new thing: Rip out your big ol' SPARC systems and downsize to Solaris x86. Save those zillions, get the promotion, and still get the warm fuzzies of a familiar and manageable enterprise platform.
Don't get me wrong: No OS other than Linux (well, maybe a couple of bsds a few years back) have crossed the threshold of my home, but (a) 15 years corporate infrastructure experience says Solaris is easier to deal with with mixed-ability staff in a massive corporate environment and (b) NEVER underestimate corporate politics and the requirement for the climber to have a ready response to the "what's your XYZ strategy, Bob?" question at the CIO's golf club. beats cost-benefit analysis every time...
Microsoft has the unparalleled advantage of a single vision driving their platform and software. Right now, the babbling bazaar that is Linux has too many voices and too many chefs to spoil the soup.
While the great many voices working on Linux insure diversity and provide a wide range of choice, I, for one, think we can benefit from just a little less chocie and a little more standardization.
Let's talk hypothetical here - Let's say Sun releases Solaris under a nice license that satisfies everybody [...] Right there, the Linux pricepoint and community support is matched.
:-)?
Linux may suck, but to me, Solaris and AIX suck much, much worse.
so if I am totally missing something, would some more-informed Linux guru clarify
You know, I can't presume to tell you what you should like.
To me, AIX's system management was a constant source of problems, their logical volume manager was a disaster, and their attempts at "improving" the UNIX linker were inept. SunOS/Solaris was even lower quality, with serious kernel and user level bugs and gaping security holes, and an utter unwillingness by Sun to address those.
I'm not sure what you see in AIX and SunOS/Solaris, but that's why they make all kinds. I was a SunOS/Solaris user for 15 years and an AIX user for 5 and you couldn't pay me enough to go back. I would hope to be able to use something nicer than Linux at some point, but so far, sadly, there is nothing that's better out there (no, Darwin doesn't cut it and neither did BeOS either; Windows NT seems hell-bent on repeating the mistakes of systems that even predate UNIX; Plan9 showed some promise, but it hasn't caught on so far).
I think a good portion of the growth of Linux has been due to IBM's very successful push to get users to run Linux on IBM's big iron AS/400 and S/390 machines for large-scale computing needs.
:-)
Mind you, I think that's a good thing because IBM gets to sell and/or lease out a lot more hardware in the long run.
I think you completely missed that a good portion of the GNU tools can be replaced by a BSD userland and it is still Linux. Also, the big guys, like KDE, Gnome, and Mozilla, have nothing to do with GNU, at all. Anyone that thinks that Linux should be called GNU/Linux really deserves to be taken out and beaten for trying to give credit to a small group that adds *some* of the userland to Linux.
Personally, I always call the versions of Linux by the distribution, as the packaging done by the distribution has much more bearing on what goes into each than GNU ever does. If I am running a Red Hat box, then it is Red Hat Linux. A Gentoo box is Gentoo Linux. While Red Hat or Gentoo may use the GNU userland, they also include parts written by the distribution itself, along with parts written by the Gnome Foundation, KDE Project, Mozilla, OpenOffice.org, etc.
I mean, by your account, a Gentoo installation would have to be called Gentoo GNU/Gnome/KDE/Mozilla/Linux.