Slashdot Mirror
Linux To Take Over Microsoft In Enterprises
shougyin writes "For years, Linux has enjoyed much of its success as a replacement for Unix. Companies turned to Linux to replace Unix servers, or for new deployments within a Unix-heavy environment. Linux is still king there, but it's starting to encroach on Microsoft as well. Big companies are planning overwhelmingly (76.4%) to add more Linux servers in the next year, and less than half (41.2%) of the companies are planning to add Windows servers in the next year. Even more interesting, nearly half (43.6%) are actively planning to decrease use of Windows servers in the next year."
For a second, I read that as "Linus To Take Over Microsoft".
I know quite a few companies who run 3-4 Windows servers for ActiveDirectory domain controllers and a lot of Linux servers as AD clients.
Once Samba4 is released, these Windows servers could be replaced as well.
This week, bogus statistics pushing an increasingly boring anti-microsoft zealotry and a pro-"operating system that takes at least one more step than windows to run any popular application or game" agenda.
I agree. Percentages are tossed around without any evidence or explanation as to how these figures were arrived at. Who was surveyed? What industries were they in? Why are they planning to add Linux servers? What function will these servers have? Why aren't they planning on adding Windows servers?
This survey is not statistically representative by all means. It is done amidst users that already use Linux and done by a Linux advocacy. I am no MSFT fan. I have not had a Windows machine in my house since 1997 (and even that was Win 3.x running under OS2 Warp). However, the reality is not as rosy as this survey would like us to see.
First of all, the majority of Windows users are SMEs and they are Windows _ONLY_. They _WILL_ buy more of the same and that is a definite. A lot of the rest is desktop estate and its essential dependencies - Exchange and their friends. 95% of these will be buying more of the same. There are very few successful desktop migrations to account for anything more than that. Even that will be an underestimate. 99% buying more of the same is more likely.
That leaves "enterprise" backend use which is pretty much what this survey is about. There is a lively migration racket going on there nowdays as most of this runs in the form of Java and friends on top of middleware stacks. Every 1-2 years the latest and greatest backend idea comes along with its migration programme. As a result servers and stacks get chucked out and replaced by others.
There Linux is gaining and the numbers are about right. However that is a very small portion of the market and misrepresenting it for the whole market is to the very best disingenuous. Additionally, it also completely ignores the "Elephant In The Corner of The Room". The merger of Sun and Oracle has created a vertical stack which will once again effectively compete for their place under the sun (pun intended) in the server room. Any stats regarding enterprise migration that assign (Sn)Oracle a negative year on year growth are frankly wishful thinking.
Baker's Law: Misery no longer loves company. Nowadays it insists on it
http://www.sigsegv.cx/
Well, I just switched my home server from Windows to Linux this very week-end. That cannot be a coincidence, right?
Write boring code, not shiny code!
wake me up when Linux starts taking over Microsoft in Desktops.
I'm happy about it, but not surprised. As the old generation of IT admins go away, newer ones are more flexible and have ways of saving money without MS in the equation. Linux is not the only solution, but one competitive alternative. Different is the Desktop, partially because it is not baked up big companies like the kernel and enterprise tools are. Canonical is an exception, but sadly a more or less lonely one.
I thought it was very funny to see 41% called "less than half", and 44% called "almost half! :D
Technically correct and true, yes, but I smell bias...
I could see both big and small companies reducing their amount of microsoft servers in the future for a couple of reasons.
1) They are joining their BPOS cloud services and therefore have less need for their own in house MS production servers. Large % of big business is joining the cloud.
2) The new server topology for exchange requires whole new separate servers or hyperv virtual servers for edge (either way its a separate server license) in addition to their CAS, hub transport, mailbox servers, etc.
Who was surveyed?
from the TFA:
the organizations surveyed were picked by the Linux Foundation End User Council
Next up:
10 out of 10 randomly selected stock brokers want more deregulation of the financial system
10 out of 10 randomly selected Taliban fighters don't trust the USA
Reading slashdot one-liner: (irm http://rss.slashdot.org/Slashdot/slashdot).rdf.item | fl title,desc*
Don Marti tears the methodology and the point of the whole survey to pieces: http://zgp.org/~dmarti/business/hands-up-who-likes-me/
This sort of surveys may have value but used like this they're just embarrassing.
This is directly from the report. "This survey was conducted with members of The Linux Foundation’s End User Council, as well as other end users identified by The Linux Foundation and Yeoman Technologies. This report is being published at The Linux Foundation End User Summit, where many respondents will be in attendance. These companies include Morgan Stanley, Goldman Sachs, Bank of America, Bristol- Myers Squibb, NTT, Deutsche Bank, Dreamworks, ADP, McKinsey and Company, Bank of New York, Barclays Capital, AIG, the US Department of Defense, MetLife, CME Group, NASDAQ QMX, the New York Stock Exchange, Goodrich, and many more."
I do not trust such assessments as much I do not trust assessments which point in the opposite direction. As much as I would like to see OS prevail CS, I do not believe this will happen any time soon or even in the distant future (under the assumption that our economic regime will not change).
Anyway, a major show stopper for small business to convert to Linux-based infrastructures is the SBS from Microsoft. Small companies have as a service infrastructure these SBS servers, which provide a mail directory service, calendars, address books. It provides web based access to these services as well as an Outlook integration. And it comes with share-point, which is also a requirement. And finally it works with all these smartphones, especially Blackberries and iPhones.
Therefore a migration effort has to take into account that the same functionality has to be provided with better QoS. While better QoS ist not the problem, the same functionality is a serious problem. Especially when it comes to more detailed properties.
But even worse, migration cannot be done in an overnight attempt. These always fail and in the end you loose a customer and they switch to MS for the rest of their lives. Therefore you need a soft migration strategy. And this is the key problem here.
While you can provide most features with lets say egroupware (which is not such a good idea, a servlet based approach would be better) you still need IMAP (dovecot), SMTP (postfix) and LDAP to model the mail service. Egroupware can also provide these calendars. But how do you replace Sharepoint? And especially how do you integrate with Sharepoint? While you switch to webdav oder sftp etc. the client's clients will not switch (at the same time). So you still need to integrate both services.
I have not seen any generic strategy for this problem. And honestly there are hundreds of thousands of small companies using SBS. And bigger companies use similar services.And the Blackberry-integration into a replacement infrastructure is very important as all these business guys use it.
Interesting enough, I was surprised to learn that some rather large institutions run their programs inside cygwin inside windows.... So all they are doing is replacing the windows machine with linux instead of running the java inside of cygwin, which makes perfect sense. As to why they were running the program inside cygwin inside windows to begin with, I have no idea.
Do we have AD like single sign on at least for linux servers? No? How about clients then? No?
Seriously, how do you guys handle root password management for servers? SSH is not the real answer here, IMHO.
While Linux is certainly the much lower costs option vs. MS, the real issue should be about Security. The problem though, is that many of *nix, is the fact that since Windows is so easily cracked. And once cracked, they have access to SSH keys and/or passwords and the ability to place a snooper. Once you have access to being on ANY TYPE BOX, it is over. It is simply a matter of time before it is fully owned. This does not matter if it is windows, Linux, OSX, trusted Linux, trusted Solaris/AIX/HP-UX, or even a os/390.
I prefer the "u" in honour as it seems to be missing these days.
RedHat support cost is killing the opportunity to increase linux in enterprises... Windows licenses are cheaper!
Linux supports rwx/ugo file permissions, as well as ACLs. It really isn't a problem.
The REASON that ACLs really aren't used much is that they are too difficult to audit. The specific problem in your referenced article can be solved with links.
Just another "Cubible(sic) Joe" 2 17 3061
From what I have seen.. the "decrease" in windows installs is because of data center consolidation and closing of offices that had a BDC. Yes more companies are looking at linux solutions for the back office, But it's not the picture they paint.
Windows is losing simply because of scaling. All companies are scaling back and if they reduce the number of servers at satellite locations, those are expensive licenses they will not have to pay.
Granted, I personally think it's retarded as hell to shrink your network like that and remove BDC's.. I experienced that at AT&T in the early 2000's we removed BDC's from offices that had less than 1000 employees. office downtime went up because when T1's to the nearest divisional office went down, productivity at that office usually took a crap. a small BDC is cheap and can serve as the office print server as well as file storage. but no, all that moves from local to at the end of a T1 or a T3 and now everything is slow as hell. Every try to support 10 users on citrix over a T1? It's painful for everyone involved.
Do not look at laser with remaining good eye.
Thanks! That's the point actually.
Write boring code, not shiny code!
I'm just trying to think of some area of tech/i.t./communications where MS is increasing it's sales ...
... Still thinking!
America, Home of the Brave.
"Would you care to comment on whether Samba4 is useful only for replicating MS technologies in the network, or also for use in a pure Linux/POSIX environment (UNIX, Linux, Mac)?"
It's certainly useful. I'm using it in almost Linux-only environment.
"Can you use pure Kerberos (not the MS version), or is that recommended?"
Yes. It's possible to use Samba4 as a pure Kerberos server, and it works just fine. In fact, I've first installed OpenLDAP+Kerberos and then migrated everything on this test site to Samba4.
A piece of trivia: it's actually possible to join WinXP into a pure Kerberos domain.
"And can Linux Terminal Server Project tie into this in some way (serve an appropriate terminal image based on a Samba profile)?"
No idea. Though I imagine that it should work.
As a poll taken by the Linux Foundation based on the answers of two hundred of it's largest members that responded, what I found suprising is that less than half of them plan on increasing their use of Linux - these are the biggest supporters of Linux, and 50%+ ARE NOT PLANNING TO INCREASE THEIR USE OF LINUX!
These are Linux's biggest supporters (they joined the foundation, they replied to the survey, and they are of a certain size) - if half of them aren't increasing use of Linux, to me that is the interesting number. If 50%+ of the largest members of the Oracle Users Group said they were not going to increase use of Oracle DB that would be the story, why is the spin backwards here? Oh yeah, Linux Foundation wrote the press release, slashdot partitas it...
Ken
41% is less than half in one sentence, but 43% is nearly half in the next. I guess 42% is the hard cutoff by which we begin referring to the statistic in the affirmative vs. the negative.
Check out my lame java blog at www.javachopshop.com
weird, I work for one of the companies on the list... our web servers for customer front-end (not business front-end) run Linux and a very small amount of our analytics run Linux... Most of our Windows servers are virtualized, so maybe that is where the numbers are going askew... My department alone added something like 2000 Windows servers this year, and about 5 new Linux -- I would estimate about 50 of those windows servers were non-virtual. Now our existing Linux servers are upgraded significantly beyond what our windows servers run (more processors, more RAM), and in general we upgrade and patch Linux servers, but we never reconfigure hardware on windows servers -- we always replace/rebuild them (company standard policy), which really adds to the "new windows machines" numbers and makes TFA's statistic seem more and more made-up. I do not have exact stats for company-wide, but I can say the support team, the internal user base, the server footprint, the development staff, the administration staff are all orders of magnitude bigger for Wintel vs Linux/Unix/iSeries(which is one team)... In other words, I do not believe this survey, based on inside information about one of the companies listed.
did it occur to you that someone has to be (in) the minority?
i didn't read the entire article nor do i have any insider info like you do but if you take any survey and the report on the findings of said survey does not describe you then obviously you were not what the survey found to be the trend that emerged from the sample group.
$ unzip, strip, touch, finger, grep, mount, fsck, more, yes,fsck,fsck,fsck,umount, sleep
Considering that the latest workstation distributions are still plagued by basic install issues (blank screens, etc), it's still not ready. I got a chuckle reading some of the support forums and the responses that were being given to 'newbies' as if they really expected these folks to understand what was being told to them. Although it has advanced by great strides in the last few years, it's still not ready. The basic install should work universally across a wide range of hardware. Seems they are still having problems with improper ATI/nVidia drivers being selected. This seems to be a common theme for the last 4 or 5 major releases.
If you install Windows, the chances are very rare that you won't get a desktop after the install. Even less so for OS X. End users don't want to deal with an OS that takes troubleshooting and 'technical' help to get it to even show something other than a command shell.
They newest offerings are impressive, but still need a LOT of work.
I think that an OS that spreads across both areas but reflects poorly in one (workstations), can effect it's perceived quality in the other arena (servers).