Slashdot Mirror
Reducing the TCO of IT with Linux?
Bil Simser asks: "I've been asked by upper management to look at the feasibility of replacing our current Windows IT infrastructure with Linux. Basically someone has said that Linux is free so now we're off to see how free that really means. A full replacement is probably impossible, but I can see some benefits coming from selective replacement of specific technologies (e.g. application servers, web servers) that might be feasible. This is both from a cost reduction standpoint and increasing productivity when it comes to system management. I've already looked at a few studies done on TCO reduction on this and they look good so now I'm turning to the Slashdot community to see if anyone has either practical experience or informative insight into a problem like this? The objective is to determine the TCO of deploying Linux as a core part of our operational environment so what does that mean in the sense of hardware, software, middleware and management impact?"
I'll be blunt.
Most people are stupid.
Stupid people don't know linux.
Teaching Stupid people == Money
I've never been really impressed by studies that claim to produce general numbers on TCO. Things depend very much on what you do and how you do it. Any study you see could be of use to you though, they can help you do your own numbers.
Introduce linux servers when its time to exchange old ones and use linux as a replacement for things where windows dont work that well. A complete overhaul at once is probably going to cost more than its worth. There is no need to toss something that works out the window.
If you replace things as they are too old/broken you dont get the problems that arise when you rip/replace everything. A slow steady pace of replacing should keep the TCO down.
That is unless you want to rip everything out and install an iron_butt of an IBM server. In that case it can save a lot of money but the investment is pretty hefty the first year.
HTTP/1.1 400
I wonder what kind of IT manager you are, expecting us to give you a detailed answer to _such_ a generic question.
:-) Next to that, yes, we are saving a lot on our IT budget by using Linux instead of Windows. But we're just a five-person shop and we all studied informatics, so we don't shy away from tweaking our systems and we don't really care about the system our software runs on.
What does your company do? What kind of software do you run, on servers, on desktops? What hardware setups do you have? What software are your employees used to? How IT-competent are your employees - will they freak out when the "start" button looks different on their desktop, will they call support when Clippy is missing? Is retraining an issue or do you use custom-made software that can be ported to the new environment? Can you estimate the cost of porting your custom-made software?
Etc. etc. etc.
Despite what the marketing people tell us, TCO is always a subjective calculcation, there is _no_ objective way of measuring it. Ask two people in the same company for a TCO calculation and you'll get massively different numbers.
Speaking from my own experience, I can say that using Linux instead of Windows has massively reduced my frustration with server setups and networked clients. A non-frustrated, happy IT manager is good for the company, so that alone should be something to consider.
------------------
You may like my a cappella music
Fact 4 - MSCE = skill shortage
Humorous signatures are over-rated.
the best I can recommend is to start small. Let everyone know that you'll be making some changes. the fisrt step is openoffice/mozilla.
the training for that will be relatively little. go one office at a time; first IT, then accounting, etc.
Use that as a yardstick. gauge the companies relative stupidity/oblivious user.
Mozilla will be easy. it might be as simple as sending a tech traininer to each dept's next meeting and saying "we're upgrading IE, it will look a little different, but it's almost the same.(use the modern skin- people tend to think of it as more of an 'upgraded' look)
next try openoffice. this will be your key. it will require retraining stupid people. This means you taking the time to document it and create a FAQ and a 'how do i...' list.
if they can make it this far with relatively little pain, then try converting a few company servers to linux(webservers are a good start.) then try the IT dept. measure how difficult it is for each person. figure out the basics of exactly how long it takes the fairly tech-saavy people to get it. then take it one office at a time.
start small. baby steps.
Looking for Book Reviews? Check out Literary Escapism.
One of the ways in which Linux + free software can help is in the removal of need to count licenses and also do the whole purchase order dance whenver adding a node or an application to an existing node. Not spending time on that stuff can be a cost savings in itself.
'If I stand on a hill in a storm with copper armour and a sword held up at arms length shouting "All the gods are bastards", will I get hit by lightning?'
(With apologies to PTerry)
Seriously - lets not retread the same old stuff on Slashdot that most of us could write code to generate the resulting opinions and flamewares (Linux vs M$ / P2P vs RIAA / SCSI vs IDE / MAC vs PC)
The best discusions I've seen here are where we get a good spread of opion, those are interesting and challenging.
Oh wait, I forgot where I was...
The majority of Ask Slashdot submissions are just plain stupid, and I have come to expect this. However, this Ask Slashdot is among one of the most boring ones that I have ever seen, followed by the most uninteresting string of comments ever.
While a discussion of the TCO of Linux may be valid, it isn't necessarily news for nerds because most nerds probably don't care about TCO unless it involves them using Linux instead of Windows at their workplace.
And, lastly, how is anyone supposed to write a comment for this that is actually funny? Everyone seems to struggle enough with this even for the more interesting stories.
>But at the same time she is very calm and composed
:)
>on a surgery table even in an emergency. I have
>seen her being at her best. But not once have I
>seen her calling others stupid, even if that person
>has no sense of medicine.
Oh I dunno
Medical people can see a lot of luserish behaviour in a typical day. It depends on where you work, geriatrics probably isn't so bad, but ER folks see every kind of stupidity you could imagine, and several you couldn't.
There are mailing lists etc. where EMTs and ER people trade stories about the outrageous stupidity of their patients, these lists have the same tone as tech people telling "CDROMs as cupholders" stories.
Ever notice that people who injure themselves because of drugs/alcohol receive a slightly different standard of care at hospitals? You've just used up all your sympathy points, dude!
Good advice to consider the overall timeframe for costs and when savings will be realized.
There are some costs that are harder to measure. Costs of unreliability, people futzing with unfamiliar or non-intuitive software, etc.
I think probably the single largest impediment to Linux adoption currently is not that the applications aren't there, it's training costs to get users up to speed with the alternatives, and it's investment in legacy "standards" - someone's got a pile of old Excel spreadsheets with important information in them.
A lot of people get overly excited at the outset: "Linux is free, reliable, open source and Microsoft charges me every which way to Sunday for stuff with bugs and security holes - no brainer dude!"
Well, training costs and investment in existing systems are real.
If I were you, I'd suggest a small prototype deployment to give you a better handle on how much the training costs will be, since they depend on your users and your mix of applications. Some users and apps will take to Linux like a duck to water, but others will bump up against stumbling blocks that might not have foreseen.
You can win with Linux (my organization is), but it's not a brainless transition.
"Provided by the management for your protection."
I would never use an tool such as kixtart... it's horribly insecure, has been caught as a virus spreader in our office and is very limited... I use Perl and a simple verification script (and server) to not only move out resources, but to schedule scripted updates in the middle of the night running as a service. (something that cant be done with kixtart) I drop a desired action script to the server and it runs on ALL machines that night, reports sucess/failure of each machine and flags on that machine that it has ran sucessfully, then will not run it again.(even version checking of my scripts so local scripts like user share mapping scripts will auto propagate from the server.)
BUT, I still need the ability for users to map printers and shares at will. Basic printer useage requires that the user can pick what printer they want (important when you have 60 printers in your office with another 10 special use printers) most users do not need access to the dye sublimation printer or the 36 inch wide inkjet printer but those that do can easily map it themselves.
it's all giving the users what they need..
Do not look at laser with remaining good eye.
Most office workers barely make use of Outlook's full complement of abilities.
Unfortunately, its not 'most office workers' you have to worry about - it's the 'muckity-mucks' who have offices, that are wed to thier shared calendaring, public folders and global address book that matter. They make the decisions, hence Exchange lives on.
You are right in saying that most people barely make use of Exchange, but the ones that do tend to be the decision makers.
I'd give my left arm to get rid of Exchange here at work. I've even suggested it, which soliceted strange looks and lots of grumbling. *sigh*
"The natural progress of things is for liberty to yield and government to gain ground." - Thomas Jefferson
If your network is fairly homogenous, the most expensive part will be getting the first couple of machines installed and configured. If you're clever about how you do the first few instances, setting up the rest will be (nearly) trivial. There aren't many cases where you'll find reasons to limit yourself to server-only replacements. Linux is capable as a desktop OS, and is much easier to administer than Windows.
IME, getting servers installed and configured is easy. Getting desktops configured is harder, because the focus software tends to be less robust. Getting Wine and various Windows apps installed; making sure the plugins for the browser(s) are installed and working; setting up the default organization desktop with app icons and such; getting the login authentication mechanism configured properly; making sure network printing works... this is the labor-intensive stuff. Again, once you get everything configured properly for the first machine, you can usually clone the configurations to new installs, so all of the work is up-front.
After that, maintenance is fairly easy if you choose the right distribution. Some are better than others in that respect. Actual sysadmin effort tends to grow logarithmically -- rather than linearly -- with the number of machines being supported (again, if they're homogeneous). Help desk support needs are about the same as for any other OS.
Where you'll find the most savings is in licensing and sysadmin costs. If you have heterogeneous hardware, sysadmin costs can go up, although (again) it is the initial installation and configuration that will hurt the most.
What some of the above posters haven't grasped is that you're being asked for a business case, not a 'Is Linux technically better than MS?' paper.
If you're being judged by business people, you need to speak their language, which all boils down to $CURRENCY_UNITS.
While you may have a template to work off if the upgrade to Win2k or NT was properly planned (and if you don't have it, it might be worth retrospectively doing this), here are the steps you'll need to go through in your business case:
The money bit
Now it gets really interesting. Assuming that you'll be calculating the costs/benefits over a number of years to produce a programme budget and calculating a break-even point some time in the future, you'll need to take into account that you're using money which would otherwise produce a return doing something else, and also that there will be inflation in the mean time.
Talk to your beancounters, and ask them what DCF rate is standard usage in the company - this is the rate by which the company assumes that money will lose its value. If you don't get one, use 10% as a fallback (but make it clear that that's your assumption). With a 10% DCF rate, a dollar will be worth a dollar today, 90 cents next year, 81 the year after, 73 in year 4 and so on - discount factors of 1, 0.9, 0.81, 0.73, 0.66 etc.
For each year, take the net operating savings (ie leaving out the initial project investment) that Linux will bring and multiply that by that year's discount factor. This will be the savings at Net Present Value (NPV - a term all beancounters consider as the real value). Keep a cumulative total.
Divide the cumulative NPV value by the programme cost of the change. This is your Return on Investment (RoI), expressed as a ratio or a percentage. When/if it reaches 1:1, you've hit breakeven. Be very clear about when you expect to hit this point - when it comes will largely determine whether you get the go-ahead.
The only thing you can accurately describe as "Scotch" is a sticky tape made by 3M. And it's
Throughout all of these posts, there are always quotes like "...now we don't have to deal with crashing machines all the time...". What in the hell are these people doing to crash their W2K boxes? It's pretty widely known that W2K is *very* stable. My boxes (I know, anecdotal evidence) don't crash. Period. A few very strained web/db servers, a few POS machines, and a few random boxes. No crashes. Ever. What in the hell are all of these people doing to get their W2K boxes to fail? I'm really, really curious.
>the Unix admins tend to have some intelligence,
You prick, what an arrogant and ignorant statement.
While I agree with your conclusion, it is easier to go from Unix to Windows than vice versa, why don't you show some insight and realize that all Windows NT are not idiots and all Unix guys are always SO intelligent.
Give me a break.
Windows System Admins might love Exchange. I did when I administered it (prior to my current Unix Admin job).
Exchange is an incredibly powerful software program. The end users aren't aware of this power, they are only aware of their interface to it, Outlook. But, without Exchange they would not have all those cool custom designed forms and templates to use.
It is these forms and templates that are designed around individual processes that are unique to the company that make Exchange indespensible and difficult to replace. We all know that qmail, postfix, sendmail, exim, courier, etc can replace the smtp or imap components. Clever imap users can even implement public folders. But, without A LOT of custom programming, administrative headache and management heartache you cannot replace the custom forms in email. Sure, you could build a database application on the side, but integrating that into email is another animal entirely.
Thankfully, most small business don't use this feature of Exchange. But, large companies, with thousands of users, do. Just like Lotus Notes and it's databases, these forms and templates have become the only way people can do their work, their information must be put in these forms. They have no choice. They can't send in a StarOffice doc, because it won't integrate with the data everyone else submitted.
Nearly every post I have read on Slashdot that mentions Exchange have been from people who never used it, only heard about it, or think they saw the Administrative console for it once. Trust me, there is no complete solution for Exchange in the OSS world. This is being said by someone who sees Evolution as a great Outlook replacement, OpenOffice.org as an MS Office replacement, and Gentoo Linux as a MS Windows replacement on the desktop.
Exchange, when it is being used fully, cannot be replaced without a change in the way people do their work.
I agree. It may be "politically correct" to say this on Slashdot, but it's true. The Windows shops I've known (and worked in) have many more servers than correspondingly powerful Unix installations, and each Unix admin was able to ride herd on more servers than a similarly situated Windows admin.
The whole Unix mindset and toolset is incredibly admin-friendly, once you grasp the simple principles.
You are using the wrong kind of licensing, and wasting a lot of money. MS offers 2 forms of client licensing - per server and per seat. With per seat, you just need a CAL for every machine that connects to any server - one workstation, 1 CAL. Here's a link that explains it.
And yeah, you were supposed to upgrade your CALs to win2k CALs.
SONY. Because caucasians are just too damn tall.
A former co-worker and I replaced NT print servers [serving ~1400 NT4 workstations across 9 locations] with a system of 2 node HA clusters running Samba on Red Hat. The problem to which you made reference has not been an issue since samba 2.2.x [almost 2 years now...]. Versions prior to 2.2.x did not officially work for NT but we managed to get it to work [much hacking of PPD files and what not...] well enough to convince our boss to let us do it.
Our final configuration was Samba 2.2.6pre2 [who's afraid of running beta code in a production environment?]on Red Hat 7.2 with CUPS backending the print jobs via IPP to the HPs. We authenicated directly against the NT4 PDC via winbind [a relatively recent addition to samba]. We had no serious problems [other than political].
Then we 'merged' with a larger company, my co-worker became by former co-worker in the first Abitrary Staff Reduction Day, control of print servers was given to a different department and they immediately moved us to Win2k print servers...
Samba 2.2.x can also act as a domain controller for a Windows NT domain. There are currently some issues with joining a 2.2.x samba server to a native win2k domain [I've been out of the loop on the details...] but I gather that the upcoming release of version 3.0 will bring samba in the the win2k/Active Directory world...
My next battle will be to get OpenOffice on some desktops instead of m$ office...