Slashdot Mirror
Windows Servers Neck and Neck with Unix Servers
BrainSurgeon writes "According to the Register, Windows based servers are now even with Unix based servers in terms of sales for the first time ever." From the article: "In an overall up server market, IDC counted $4.2bn worth of Microsoft Windows server sales on the back of 12 percent growth. Total Unix sales also hit $4.2bn in the period, IDC said, on 3 per cent revenue growth. Those totals left Microsoft and Unix systems holding 35 per cent of the server market each."
What the heck is running the other 30%?
Netware and OS/X?
See my blog http://ilovecookes.blogspot.com/ for light hearted technical information.
Who wants to bet that maybe Microsoft just charge more? :)
the layman's guide to computer science
The creatures outside looked from Alt-Right to Antifa; but already it was impossible to say which was which.
I'm suprised they didnt mention that Linux servers had the greatest overall growth with 35.4%, and that they're 10% of the entire market. Now if Microsoft hits the 50% mark then thats when i'll start believingthe whole Unix/*BSD is dead hype
drunk chemists
That doesn't count towards how many servers are running linux/unix without having paid for it... in regards to how many servers are out there of each, you can't go by just the sales. I would say linux/unix probably outweighs windows in quantity of servers on the net.
You're nothing; like me.
I wonder if this will make UNIX vendors realize that unless they can get a consumer UNIX, like Linux, out there and viable, erosion is going to continue. Windows will continue to eat up the server space for as long as people need Exchange and Active Directory servers. We need to remove the need for Exchange and Active Directory servers.
Does this mean they took the value of the entire server market and calculated the share of Windows server from the size of the sales... (100/35)*4.2bn = $12bn total market? Doesn't this mean that the amount of Unix installs per server is still much bigger because Unix (if it includes *bsd) is cheaper?
blade servers are VERY hot, at least compared to the density. A full rack of blades is nearly impossible, you simply can't cool the rack enough to run it. (most vendors don't tell you this, although IBM recently started admitting that fact)
I wonder if this isn't more of a sign that OSS is making some headway. Linux server sales are way up according to the article, and they compare Unix and Windows servers based on cost.
My understanding is that more major server sales folks who are pushing some Unix flavor are trying to make their money on the Service that goes with the server, not the actual initial sale. In which case it would make sense that you could knock the price down on the Unix server that's running a free OS vs. the same machine that has a 500 CAL license for Windows 2003.
I wish they would have given us number of units vs. the cost of units.
This is just murky adspeak.
10 idiots want to buy a lemon for $10/each and 100 people buy a tasty pear for $1/each.
Is it just me or does it seem like there are still a hell of lot more pears out there...
It is not neck and neck anymore, I just brought a windows server so Windows wins by massive margin of US$4,000.
so with the same number of servers, the load they are processing is 50-50 too, right?
.0001 .0002
Here are some official numbers from the WGAF 2005 Study of total workload being handled on the net.
Unix 85%
Windows 10%
Other 5%
HOWEVER, these numbers get funny when you factor in computing time spend running malware.
Windows outperforms Unix and others in this important category. Truly Windows has no peers when executing excrement.
Unix
Other
Windows 99.9997
These are REAL NUMBERS folks. every one of them. even the 7
So in this case the sales increase is not necessarily based on the quality of the offering but on the convenience.
The report actually indicates that Windows Servers are gaining a smaller share of the server market INCREASE than they should, and Linux is gaining TWICE as much as it should if they were all actually gaining an equal share.
Also a number of idiot commentators are saying "Windows servers wipes the floor with Linux" when in fact the report shows that both Windows and Linux are wiping the floor with PROPRIETARY UNIX.
Yawn - big surprise. This has been a foregone conclusion of every analyst for the past two or three years - that Linux (and to a lesser degree Windows) will replace proprietary UNIX and then the battle will come down to Linux vrs Windows - which Linux will win handily.
Richard Steven Hack - This sig is TOO GODDAMN SHORT TO DO ANYTHING USEFUL WITH! MORONS!
Mostly big "enterprise" CRM and other slaes type applications, as well as document management systems. And of course IIS...
"Who are in control, they are not in control of anything - they don't even control themselves!" - Glen Beck
As is always the case...
The raw number of sales between Unix-based and Microsoft-based servers not being considered by the article. The dollar-value of sales is what they're looking at. In terms of dollar value, as much money was spent on Microsoft-based servers as on Unix-based servers, at $4.2bn
If you're going to talk about the real number of servers being implemented, you need to consider the fact that, in general, Microsoft-based solutions cost a whole lot more than Unix-based solutions.
Interestingly enough, $1.2bn was spent on Linux-based servers, and Linux-based servers accounted for the largest increase in sales.
If you believe everything you read, you'd better not read. - Japanese proverb
One of the reasons Microsoft is making such inroads into the server market is that they've really improved their operating system. Windows servers can be made reliable and secure if they're administered properly. Insofar as it brings choice to the marketplace, then having Windows as a realistic option for a server is a good thing.
But don't lose sight of what's at stake. The Microsoft business model is to leverage it's monopoly in one area to drive out competition in another. If Microsoft will let Windows coexist peacefully with it's neighbors, then great. If they're true to form, though, they'll introduce incompatabilities and do everything they can to make sure businesses don't have any more of a choice in their server OS than their desktop OS.
The struggle isn't just about running the cooler OS, or using the command line vs. a GUI. It's about freedom and choice.
KTHXBYE
Do blade servers run hot? Or do the British heat their knives over an open flame? Can someone explain to me what the pun is in this sentence???
Maybe he's just been doing hot knives.
Give me Classic Slashdot or give me death!
I guess my home web server doesn't count in the statistics for server sales in the past year. I found the machine on the street, threw in a new hard drive from newegg, and it's running Debian GNU/Linux. I just did some work for a company that wanted me to get rid of their spam and also upgrade their website backend. The website backend project involved setting up mysql on an old server that they had laying around, and the spam filter is postfix+spamassassin+amavis+clamav running on Debian on an extra desktop workstation computer that they had at the office. I also set up a near-identical spam filtering machine using Debian on a workstation at a law firm recently. So, there you go, 4 new servers running Debian GNU/linux in the past year. It's interesting that these sales figures are dollar denominated. It marginalizes GNU/Linux as a server OS (because it's free as in beer or alternately relatively cheap if it comes with support) while also informing how valuable it is (because in the hands of the skilled, it gets the job done while costing nothing).
Uh, interesting way to spin the original press release, which prominently highlights 35 percent revenue growth for Linux, 12 percent Windows, 3 percent "Unix," which should really be called "Other Unix."
Granted, the Linux server $1.2 billion factory revenue is less than a third of the Unix and less than a third of the Windows market, but hardly insignificant. Also much harder to trace, I reckon, given how many people strip Windows off a Dell and make a Linux server with a spare copy of Debian.
Linux usage is certainly up, the ROLE of the server needs to be discussed. I have 24 linux machines at one client's location, too bad most of them are just proxy servers running squid with an LAMP for configuring them. While all the chatter about Linux making progress into corporate IT the role of the machine needs to be involved in the discussion?
How many corporations have Linux-run PDCs? Email? File Respositories? Backup? All this talk about sales figures means little when you take out the role of the server out of the discussion. Without a breakdown along the Lines of X Windows 2003 Email Servers vs. Y Linux Email Servers the discussion really has little value besides a vague sales figure. The discussion of Linux, BSD, Windows, BEOS, Tiger, whatever is is lacking any real worth. Going on 11 years here soon and corporations are not cut and dry. What does this follow fact tell you (taken from one of my clients):
# of Linux Machines 3
# Of Windows 2003 Servers 24
# Of Windows 2000 Professional Machines 8
What do they use more? Windows? Not really. The 24 2003 Servers are used to simulate web traffic and other customized in-house traffic. Not one of those Windows servers is mission critical. The 8 2000 machines are the staff's workstations. The core critical machine that run's their entire manufacturing system is a linux machine. 1 Linux email server, and 1 linux firewall. Now looking at that figure you couldn't determine how important any of those servers are, we need more data in these discussions, it's incomplete.
Purchasing numbers mean little. Even across a broad scope there is no direct correlation between number of copies of X and their level of importance in a company, if you think that probaility shows that given there are 2 milion copies of A and only 1 million copies B that A is used more in mission critical services I would recomend you avoid gambling. The Christian Bible is in over 50% of homes yet less then 10% of people can repeat the opening of Genesis. ("In the Beginning God created the heavens and the Earth" I believe.)
I'll summarize with a classic Ken-ism:
OWNERSHIP OF SOMETHING DOES NOT GARUNTEE THE UTILIZATION OF SOMETHING.
Take a typical computer, slap some SNMP on it and grab CACTI and monitor the staff in a building. I bet you the average work-hour utilization of the processors will never exceed 50%. 10 years is hasn't. Just because you have a 3.4 GHz processor doesn't mean you'll use all that CPU power.
For you drivers out there your speedometer can post 125 MPH... doesn't mean your gonna ever go that fast right?
--END RANT + LUV--
-=[ Who Is John Galt? ]=-
An example I recently was involved in: I work with a company doing software. Big and mission-critical systems. One big customer wanted a really big installation of this software. We recommended that this was run on either Linux, BSD or Solaris. Our customer had hired their own consultancy company, and these consultants were very pro-Microsoft. So the customer said "We need this to run on MS-Windows", and we said "Ok, our software can run on MS-Windows, although we cannot recommend it.".
So a big server park was ordered with MS-Windows preinstalled.
Then, as the project progressed, the customer also hired an Oracle consultant. This consultant said "I would not sleep well at night if these Oracle servers are running on MS-Windows. Other systems will give you more stable operation". So all the operating systems on the Oracle servers were scrapped, and Linux was installed instead.
Then, when all the servers were sent to a hosting provider, the hosting provider said to the customer "We see that while the Oracle servers run Linux, all the application servers run MS-Windows. We will be better at supporting this system if all the servers run the same OS, and you will probably have better uptime if running linux on the application servers too. If you don't mind we will install Linux on the application servers for you free of charge.". The customer accepted.
So while this big server park was purchased with MS-Windows pre-installed, all servers were running Linux before the system was put into use.
Be on the lookout for web services that are up for a few days, get /.'ed, and go poof. You can thank Gates for that.
No, you can thank poor administration and low bandwidth for that. Default Apache isn't going to stand up to a Slashdotting more than IIS. Would I run IIS? No. But that's besides the point.
"Nature doesn't care how smart you are. You can still be wrong." - Richard Feynman
Now there's a bad start.
My Linux Command of the Day site : LCOD
Adding to your comment, it doesn't figure how many "obsolete" servers are running Linux either. My old Dell P3 Server is too old to run Windows anything, but works nicely as a firewall. (p3/1ghz/256mb) More than enough power after being retired from being a Windows server.
It could easily serve over 150 users of static web content (probably over 500 users if the content is truly static). I still have an 1997 IBM PC325, dual PPro 200mhz w/384mb ram that can't possibly run any version of windows, but it has an uptime over 300 days, and linux installed on it since 1999 (new version two years ago) and is STILL a great primary DNS server. (actually I have a few of these doing various task). The stats do not reflect this.
I use old hardware for routers, dns, irc servers, test servers, honeypots, and non critical web servers all the time. These stats do not reflect ANY of this. And I pay for support from local companies to help reduce costs, but don't "pay" for Linux. All the servers run Linux now.
I also buy shiney new dual Xeon servers with RAID systems and gigs of ram, but I don't buy any OS, I just use Linux. I guess those aren't counted either.
All my Linux desktops started their life as "Windows Preinstalled" as well. Ironically, it was cheaper to get them with Windows than without.
Tequila: It's not just for breakfast anymore!
Excellent point.
Microsoft explicitly recommends running each server app on a DIFFERENT server. Don't run your Exchange on your Active Directory; don't run your license compliance app on the same server; don't run your SQL Server database on the same server.
Why? Simply because MS server app performance sucks because of "featuritis bloatware".
So how many of these new Windows servers in the study were actually running ONE-THIRD of what the Linux servers were doing?
As someone else pointed out, the roles these servers play have a bearing on the value of the figures.
Richard Steven Hack - This sig is TOO GODDAMN SHORT TO DO ANYTHING USEFUL WITH! MORONS!
Unix is more typically loaded up, running as many things as the hardware can handle. When it starts getting too loaded then you buy another one (usually a bigger one).
We've recently bought two quad processor linux machines running vmware to run a dozen or more windows servers. Two linux sales, a dozen windows sales.
Nerd: Derogatory term typically directed at anybody with a lower Slashdot ID than you.
Quick! Somebody buy another server and break the tie!
I would say, SQL Server is one of the few remaining reasons why Windows might be a better bet in some circumstances. It's really a fairly impressive database product. I suppose its closest competition on the Unix side is, well, Oracle, or PostgreSQL, but then you aren't even touching the Analysis Services datacube technology, which is fairly amazing and seems to have no peer in that industry.
How about OS/400. Lots of retail POS ISVs run the server code on AS/400 (iSeries). And how about zOS? Still a lot of banks and insurance companies run their core apps on mainframe.
Can you measure and compare what the RAS level for a server is? The Software (OS and application) and Hardware in combination would play a factor in the RAS level. I would like to see a mathematical formula based on MTBF for hardware components (especially hard drives, power supplies) and OS and application software quality quantification(1) to create a RASmark level. It would help make server buying decisions less seat-of-the-pants so you can decide whether or not it's worth it to get the redundant power supply option and/or RAID level for a server to get to a required RAS level for your needs.
(1) It's difficult but not impossible to quantify software quality. There's plenty of real-world usage that can be surveyed to cancel out admin competence levels (another difficult item to measure) and other factors. Have to beware of zealots and a certain monopolist's FUD.
Transmitting energy without a license.
Remember, these numbers are based on dollars, not number of servers. Almost all of the servers you're talking about, being sold without an operating system so Linux can be installed later, are going to be way down the line in terms of cost (and cost is both cause and effect there). The number of Linux servers is going to be significantly higher than the 10% figure given in the article. They're also probably much more efficient with their dollars, able to accomplish more for the same number of dollars (e.g. ten $1,000 Linux-based servers may be able to do a lot more than one $100,000 name-brand Unix or Windows server "solution", and for the same amount in support costs, you probably get about the same amount of service.