Slashdot Log In
Australian TCO Study: Linux Wins Again
Posted by
timothy
on Mon Dec 13, 2004 04:21 AM
from the ask-enough-and-you-shall-receive dept.
from the ask-enough-and-you-shall-receive dept.
An anonymous reader writes "An updated Linux vs Windows TCO study has found that a 250-seat company can end up saving 36 percent if it were to equip its users with the open source operating system and applications that run on it."
This discussion has been archived.
No new comments can be posted.
The Fine Print: The following comments are owned by whoever posted them. We are not responsible for them in any way.
Full
Abbreviated
Hidden
Loading... please wait.
What about a larger company (Score:4, Interesting)
Re:What about a larger company (Score:3, Funny)
Re:What about a larger company (Score:3, Funny)
The big advantages with Windows infrastructure are the tools for managing lots of machines (eg: Group Policy) and the ease of integration.
Re:What about a larger company (Score:5, Informative)
Only if you haven't used Unix extensively. Compared to Windows managing multiple computers in Unix/Linux is trivial. You scripts don't care how many computers they connect to after all.
And managing things like AV/Firewall/WindowsUpdate is still not as streamlined as it can be on a Unix system.
Parent
Identify functionality, not products. (Score:3, Insightful)
Don't just rattle off names.
Identify the functionality that each of those provides and WHY it is necessary for an administrator.
Only then can you compare/contrast the two platforms.
Re:Identify functionality, not products. (Score:3, Informative)
Of course there are also tools like radmind too.
But the most significant thing is that you use linux differently then you use windows. For example you may use thin clients which eliminate the need to manage desktops at all. You may choose to mount a common
Unix was designed from day one to be mass managed. Tha
Re:What about a larger company (Score:3, Informative)
On one machine, using software like XGrid [apple.com], select all the machines you want to update, and issue the command. Sit back and watch as all machines you're updating complete their task.
Both Windows and Linux are now at the stage with automatic updates where large organisation can have one machine that downloads the updates, and acts as a server for the rest to get their updates from. This is as good as it gets, and both systems are there already. I'd be surprised if the same isn't true for updating of
Re:What about a larger company (Score:3, Informative)
The separate question which you seem to be asking is whether that's still true accounting for employee turnover. Well, I've not done any study on it myself, but if you're going to bring up retraining of new employees then you also have to consider the continued year-on-year savings of not
Re:What about a larger company (Score:3, Insightful)
However, after to speaking with a few of the higher up IT guys at various trade shows and other events where we accidentally windup in the same room. I have concluded money has very little to do with us using Microsoft products. Rather it's other things like: PHB's (almost by definition) aren't highly technical people but maintainers of the status quo, "No one ever got fired for buying Microsoft", and most
Re:What about - so why not IBM still? (Score:3, Insightful)
Re:What about a larger company (Score:4, Insightful)
I remember when that saying went "No one ever got fired for buying IBM" and it's really not that long ago... Things change. Always have, always will.
Parent
Re:Screw TCO (Score:4, Interesting)
We're talking about big multinational companies, so a lot of your evangelical strategies won't work, are inappropriate, aren't welcome and would get you fired. For example:
In Step One: I work in R&D and my Boss (in fact the whole food chain from me up) is a Ph.D. Physical Chemist, and despite the fact and he's got the message (he uses firefox at home, for example) he has no control (or interest) over what IT does and thus I would be preaching to choir. All of the desktops in the company are standardized (choice of 4 types) and locked down, no one has write privileges to the local drives or local admin rights. Running an application that is not approved is a fireable offense, So is modifying the registry, Running a P2P app, Running a server, and Bypassing security. Setting up and running a wireless network will result in the IT guys immediately, on discovery (random 802.11x sweeps), escort you out of the building. Need something different or package installed? It's no problem, but you can't do it, IT does it remotely.
In Step Two: Are you kidding me? They are not my servers to do anything with! I can not even enter the room they are in! They'd escort you out of the building.
Step Three... Back to the PHB thing, the head of IT does not live or work in the same country I do, he's never even been on site, there is no way I could drop anything on his desk and if I did, it would be extremely unwelcome because not only am I not in his field, he's never met me. BSA is meaningless to us, we have site licenses for Microsoft's, Adobe's, & PTC's entire portfolio (along with a pile of other's, it's a 48 page catalog) and we're big enough to say, piss off you can't come in and inspect (trade secrets, you know).
Step Four is the only thing you've said that makes sense or even vaguely doable, but it lacks a keyword: "Validated" and because of that would not considered.
So what does that leave me with? Only things in MY domain: Data Collection, Device Control, Device Firmware and Molecular modeling. Here I've done a fair job. I use SuSE linux on most of the data collection and machine control boxes. I use SuSE, Free-DOS and Win-XP to develop on. If you look under the skirts of a lot of our devices you find that only the older ones have custom kernels, while the newer ones run NetBSD or Linux.
I hope I haven't offended you or been overly negative, but a lot of OSS evangelists do NOT understand big companies. That's a large part of why we're still using Microsoft products.
Don't get me wrong, I want to believe!
Parent
Re:What about a larger company (Score:5, Informative)
If you get the CEO's backing it can be done as long as it is not rushed and your prepared to make it a long term goal. Middle management will always make things difficult, they have grown up on excel, vb etc. But as long as you have support from the top and dont stand on there toes to much it can be done.
* Start with web server, dns and dhcp migration to linux.
* Migrate the file servers to samba.
* Follow that by email.
* Replace browsers with firefox.
* Replace outlook with evolution or thunderbird.
* Start slow process of migrating desktop machines to linux. Start with upper management and people who only user email + open office. Single out a department for this if you can. X terminals can be a useful tool here.
* Look at replacing key database applications with open source alternatives. Most SQL database have unix and linux versions, expect for MS SQL.
Over a long time you can afford to look at replacing key infustructure.
* Replace ms office with open office.
* The small time custom apps that the organisation has collected over the last 20 years or any apps that are going to be too expensitve to port, place them on a w2k terminal server and access them from linux rdesktop. Over next 20 years they can be phased out.
* Complete migration to linux desktop.
* If there is an art department that use windows, use Mac OS X as your target platform.
* Leave the middle managers there windows laptops, just firewall them off. When they die or get to slow replace them with linux or powerbook laptops.
At the end try and aim for a couple dozen windows terminal servers to run whatever the organisation is still dependent on for windows, firewall these off to protect against virus and disable internet access on them. After 5 years these windows servers will slowly be decommissioned and the organisation would have made the complete switch.
Parent
Re:What about a larger company (Score:3, Funny)
The computers or the middle managers??
Here is how novell did it. (Score:3, Informative)
Get rid of ms office and install OO on windows desktops.
use NDS with windows client for your directory.
Install ifolder on windows desktops and instruct users to put all their documents in their ifolder.
Once the users are comfortable with groupwise, ifolder, and OO switch them over to linux running the same apps.
Smart and painless. The idea is to keep them on windows on the desktop until the end.
Note that products like NDS and groupwise
Re:Email migration? (Score:3, Insightful)
Of course there are commercial software too such as groupwise, lotus notes, hp openmail, bynari etc.
Exchange is no longer a barrier.
Why not compare to Mac OS??? (Score:4, Interesting)
Parent
Re:Why not compare to Mac OS??? (Score:4, Funny)
Unfortunately, there's no cost savings switching to Mac; in fact the cost to the company goes up because they end up having to remove the cheap industrial drip coffee maker and replace with a latte machine and more expensive coffee.
Parent
Probably a load of rubbish (Score:4, Insightful)
Note for slashdot bias fans: "Linux wins again" is actually in the story in the link, rather than a bit of spin on the part of everyone's favourite news site :)
Beware of spurious precision! (Score:5, Insightful)
TCO studies are just standard business cost estimation models, with assumptions chosen by the authors of the study. Most of the models are pretty good, in theory, with sound reasoning and empirically-supportable construction. If they didn't work, or if they tended to provide misleading results when applied properly, why would businesses use them at all?
The problem is with the assumptions. Give me any financial model, from cost estimation to marketing models to arbitrage scenarios, and I can plug assumptions into it that will give any result you want. The models are fine, but the results are "the pits", as it were, unless the assumptions are carefully and honestly chosen.
This isn't to say that a TCO model, even with well-chosen assumptions, can provide an incredible amount of precision, but it CAN provide accuracy of result. That's what REALLY pisses me off about this article--they're quoting numbers to a whole percent, when it's pretty obvious that the precision of the result isn't anywhere near %10. If the article is to be believed, they're using intentionally pessimistic assumptions in order to bias the study against F/OSS, and still coming out with F/OSS on top. They're acknowledging that they can't bring supportable, precise assumptions into it!
So really, the study is saying "F/OSS is cheaper than MS by a good margin, but our precision is shitty enough that our actual number doesn't mean much. It might be %37 cheaper, it might be %80 cheaper, or it might be %1 cheaper--but we're pretty sure it's cheaper."
I guess it's like that old joke, where the museum guest asks the tour guide "How old are these dinosaur bones?" The guide says "The bones are 2 million and 10 years old." The guest, astonished, exclaims "That's amazing! How can we know the age so precisely, when it's that old?"
The guide responds, "Well, it was 2 million years old when I started working here, and I've been working here for 10 years."
Parent
Re:Probably a load of rubbish (Score:3, Insightful)
Benchmarks are usually pretty unreliable and fudgeable anyway, but I think these TCO studies are the pits. I certainly don't believe them...
Ahh, but you've missed the point slightly: PHBs love statistics to "prove" things.
Most geeks already know the score, but TCO benchmarks aren't aimed at us, they're aimed at the PHBs. We can bang on about "freedom" as much as we like, but until someone has "proved" it will cost less, the PHBs won't give a damn!
Actual Study PDF (Score:4, Informative)
uh (Score:2, Funny)
Re:uh (Score:5, Interesting)
Sounds like you need to be using Firefox [mozilla.org], a free open source web browser... suitablly equipped with the Adblock extension. Then you wouldn't keep seeing the Microsoft adverts :-)
Not having to read the Microsoft adverts will therefore increase your productivity. Proof that Open Source software improves TCO!
Parent
TFA looks quite unbiased... (Score:4, Interesting)
But you underestimate the staffing issues there. Firing all your MSFT IT guys and hiring new "LinuxCompatible" admins is a big pain for most companies. Of course you fire 3 Win32 admins and hire one Linux admin by default :)
For a new startup, a Linux desktop is invaluable , especially if you have a couple of in-house developers who use it regularly. That's where linux is slowly creeping into the desktop - not in the big companies with million dollar CTOs and kickbacks from Microsoft.Re:TFA looks quite unbiased... (Score:3, Funny)
I assemble the systems staff in a meeting room. When they are all settled in I say, "Raise your hand if you are an MSCE. (pause) Ok, everybody else still has a job. Meeting adjourned."
Biased in MS Favour (Score:5, Interesting)
It is very interesting the assumptions that they state have been made to bias this report in Microsoft's favour.
Wow!
Re:Biased in MS Favour (Score:4, Insightful)
Oddly enough the high costs of external consultants is often greatly exaggerated. Unlike full time employees who need other benefits as well Health Care, Retirement,
Parent
Re:Biased in MS Favour (Score:4, Informative)
Yes, I always make sure I schedule hard drive and power supply failures well in advance so that everyone can save their work beforehand.
TWW
Parent
TCO Studies a waste of time? (Score:3, Interesting)
Tired of all this... (Score:4, Insightful)
In fact, I wish Microsoft would focus on something else. It's funny, but *cost* isn't something that seems to be a strength of MS. They should focus on their strengths (like consistent interface that everyone knows, massive hardware support, number of applications available, good multimedia support, etc). They have a lot going for them. Why do they always focus on the thing that they don't have going for them!!!!
--End rant.
Re:Tired of all this... (Score:5, Insightful)
TCO is all that matters.
Say it again kids.
TCO is all that matters.
A company makes a product. Technology is a means to an end. TCO is the TOTAL cost (in cash, lost sales, employee time, overhead) of the technology.
TCO includes: the cost to initially purchase the software, the cost in lost time as users and admins to learn new interfaces, the cost in paying employees in maintaining the system, the cost in purchasing obscure or less capable hardware supported by the technology, the cost in lost time in porting/writing/purchasing applications to run in the environment, and on and on.
TCO is NOT cost of purchase + cost of support. And it is also always an estimate because of so many variables it must encompass - that's why there are so many studies about TCO. It's an ambiguous metric.
TCO is all that matters, TCO is all that matters, TCO is all that matters.
Parent
Re:Tired of all this... (Score:3, Insightful)
Thanks for getting it. It's a totally incalculable measure, and as such, all these already biased studies are meaningless, except as marketing.
But when you tell your boss that you run Linux because it has higher uptime, you're translating from "This makes my job easier" - aka employee speak - to "This makes production cheaper" - aka management speak, also known this week as TCO.
We can change the TLA all we want, and MS, and Sun, and OSDL, and IBM, and anyone else playing the game will, whenever it su
Re:Tired of all this... (Score:3, Insightful)
Now their competition is still superior, as it always has been, but it's now cheaper too.. Microsoft can no longer offer a cheap crap solution, theyre offering an expensive crap solution but theyre trying to hold on to the advantage they used to have be
but wait (Score:5, Funny)
Seriosuly. ... (Score:3, Informative)
There was a time when 0wn3rship by spam bots were not even considered a problem because everyone was on dialup anyway. More recently with the coming of broadband and lots of stupid users to the internet - that has become the major headache (ie spyware, malware and trojans are local issues, spam bots are bigger).
It's a real cost when the ISP cuts you off or sends you a fat bandwidth bill :)
Patch Day! (Score:3, Interesting)
Now, how often would you have to do that on which OS?
linux has it's own supportability issues (Score:3, Insightful)
take setting up a new website:
"oh - there's a GUI tool for that... if you installed the right package... did you pick gnome or KDE?... X isn't starting? it might just be easier to modify the .conf file with Pico... don't have that? try vi - httpd.conf should be under /etc/httpd - unless you..."
Any idiot (like myself) can fumble through doing this stuff on Windows.
Security? Go to Windowsupdate.com once a month and install all the patches. I wish I had as straight forward a solution for my Linux boxes.
don't get me wrong - I want to see open source crush microsoft - it's just there's some significant work that needs to be done on the usability / supportability front.
Re:linux has it's own supportability issues (Score:4, Insightful)
A TCO is about all the cost. Installation, configuration, operation.
Yes, X can be a pain to start if it is not already configured. To release that job from the beginning users, many distributions now have hardware detection tools and configure X for you. I invite you to check most popular distributions that have been released since 1999.
On the other hand, I'm quite curious if you ever had to deal with a MS-Windows computer that crashed during the loading of the graphic card driver/window server/window manager. On GNU/Linux you have to use one of the many editors, surf for some references and write the proper parameters to a file. On MS-Windows, my own experience is that searching for information will mostly lead you to "my graphic card is not working" kinda-post, no extra help, that the only editor is EDIT, and that you have to be very lucky for the problem to be located in a file that EDIT can open and modify without totally destroying (ie: binaries are out of the question). Most knowledgeble MS-Windows user have an answer about this. Re-install.
Maybe its just me, but I prefer the option of 45 minutes from browsing for the information to the end of the problem, vs sitting in front of the computer for 1 hour(OS) 1 hour (Office Suite) 3 hours (archiving utility, acrobat, IM client and other favorite miscalineous utilities) watching the progress bar slowly moving.
Any idiot (like yourself) can do EXACTLY the same in GNU/Linux. Many GNU/Linux distributions target idiots just like yourself. Just to name one, Mandrake has a full set of utilities that will allow you to click your way to the configuration of your dreams.
And Webmin that will allow you to configure your machine from a browser.
And you still have access to the configuration files through text editors.
Security? make a cron job that check the security updates every night on your computer, and install them for you. You dont even need to go to some web site. You dont even need to wait a whole month to fix a hole.
Cron is too complex for you, again, just click your way to an updated system. Many distributions will inform you by email of every security update available, based on the software you have on your machine. Which mean you keep your OS _AND_ your applications up to date and bug free, rather than your OS and office suite.
Again, cron is a bit old school. I'm betting is most distribution do not offer you a clickable way to tell the update system to run at regular interval, its a matters of weeks before you see it.
I think you have listened to one too many bad opinion and are due to actually try it on your own. Go to www.distrowatch.com and get yourself a desktop distribution. I am saying desktop as you seem font of having kde/gnome and X. A desktop distribution would (Fedora/Mandrake/Suse/LInspire/many other) include hardware detection and configuration of the X server for you.
Try it up, its not longer 1999. And next time your system decide to play a trick on you, you will have an other option than watching countless progress bar as your only fix.
-ph
Parent
Size of company (Score:3, Interesting)
The biggest short term win in TCO will come when the organisation is of such a size and complexity that it really only needs 1(one) committed open systems evangelist to drive through change. What slows down change in most organisations is the fact that most of the workers (and managers...) are not hugely intelligent - even in IT - and oppose anything that involves change or learning.
If this is right, OSS will only really start to gain momentum where smaller companies which are adopters gain a competitive advantage that enables them to grow faster than the competition. Although IT is only a few percent of the business, a large saving in IT can make a considerable difference to the net profit - but it needs to be a large saving as a percentage of IT costs to make a real impact.
This is good news for call centres and bad news for heavy industry. It would be a pity if OSS is associated in most people's minds with the modern version of cotton picking rather than high tech, but that could be the outcome.
36% TCO. BFD (Score:5, Insightful)
TCO is a PHB metric. Managers who don't understand the role of technology in their organization view technology as a necessary evil and want to keep the cost as low as possible.
Before looking at TCO, managers should looks at:
- how much IT increases productivity
- how much IT cuts costs in other parts of the company
These metrics are notoriously hard to measure while TCO is mostly contained within the IT budget and so is easier to calculate. An astute office politician can claim some benefits just by reducing his IT costs while ignoring the effects on the rest of the organization.However, the big gains are outside IT. If IT offers a mere 1% increase in productivity in the organization as a whole it would dwarf any savings in IT costs. If IT isn't providing those types of benefits annually, it is doing something very wrong.
Return on investment, not TCO, is a better measurement of value. Businesses that think they can cost-cut their way to success are generally doomed anyway.
The point at which you stop taking them seriously (Score:4, Insightful)
If you get past that, the inclusion of Fedora Core 2 as an OS option should stop you in your tracks.
And if you manage to get past that, the needless use of, for example, enterprise versions of Windows 2003 Server should be the final indicator at how flawed their methods are.
Not a fit for all companies (Score:3, Interesting)
This may change with Novell's Enterprise Server comming out in January.
Central user management with single sign-on? It's a pain in the butt right now on Linux. How does that impact TCO?
What about all our apps that don't run on Linux? Speech to Text stuff that always falls apart in Wine, special educational packages that aren't supported on Linux? That doesn't help the TCO analysis either.
We've got lots of hardware that won't EVER work in linux - network scanners, copiers and printers, raid controllers, CD-burners, network fax machines...etc. This isn't really Linux's fault - it's the hardware manufacturer's fault - but the TCO problem falls squarely on "Linux". Should we pay BIG bucks to replace all that hardware so we can save a little money on the OS?
These studies aren't very good for anything except "rallying the troops".
Those MS TCO studies that claim you only need 2 or 3 guys to support 15,000 windows users all over the world are also good for a laugh as well.
-ted
See "Look at the Numbers!" for more on OSS/FS TCO (Score:3, Informative)
Re:Is that a surprise? (Score:2, Funny)
Re:Is that a surprise? (Score:2, Redundant)
Re:Is that a surprise? (Score:5, Insightful)
Linux: Type your username (if not filled in automaticly) and type your password.
Training: 10 seconds: 'This is the new login screen'
Windows: Click on some world or web or 'e' icon to get internet explorer, use urls, home, back and forward buttons.
Linux: Click on soome world or web or wathever icon to get an Firefox window, use urls, home, back and forward buttons.
Training: 20 seconds: 'Click on this icon instead of the old one (the one that says INTERNET), further browsing is the same.'
Windows: Click on the word icon and type your text, click on the excel icon and fill your sheet.
Linux: Click on the swriter icon and type your text. Click on the scalc icon and fill your sheet.
Training: again pointing out the new icons.
We just covered the training for 90% of all desktop users. They simply don't know or need more.
It gets interesting when you get to the artists or the real power users but they are generally a minority or have enough brains to figure most out themselves.
Further you can swap fileservers, dns, proxies, printservers and webservers in you company wihtout this 90% even noticing.
For this 90% training is mostly comforting them to make sure they don't panic when they hear something is going to change.
Jeroen
Parent
Re:Crap (Score:5, Interesting)
Still cheaper. You can't necessarily put numbers on the price of spyware and reboots, but whatever that number is, Linux is cheaper than it already. It is not a case of "Linux is free if your time has no value" - it's that "even if you value your time at 3 times the price that you would on Windows, you are still better off".
Parent
Re:Netcraft confirms it... (Score:5, Insightful)
But the tide is changing. IE marketshare is falling. According to some reports, about a fifth of surfers use alteranitve browsers. That gives serious reason to make websites that work with other browsers (yes, that means you, gmail).
People are increasingly eager to abandon Windows. It's funny that lately, many of my non-CS friends have started learning to work with Linux, and it's mostly the people who think they can handle their computers who stick to Windows.
Of course, there are still applications that will tie people to Windows. However, if people actually attempt to switch, they will learn which applications and file formats cause problems, and be more open to using alternatives. I've seen this happen in several places.
Now, all this is not to say that Microsoft will go down (I personally believe they will at least survive, if not prosper). However, now that their dominance is starting to slip, there are serious opportunities for competitiors to establish themselves in the market.
And they're trying. The other day, I heard a Novell ad advocating open source on the radio. Even if they are the only one now, where one leads, others will follow.
What would really kill Microsoft's deathgrip would be if a competitor not only did the same things better, but also offered features that Microsoft doesn't. Two examples would be efficient use of metadata (a la BeOS; this is being worked on by all camps) and truly interactive web applications (like XAML promises; Java and XUL are just not good enough).
Parent
Re:Netcraft confirms it... (Score:3, Insightful)