Slashdot Mirror
Ret. World Bank CTO on Desktop Linux TCO Facts
comforteagle writes "W. McDonald Buck, retired CTO of World Bank, believes we need to take a more honest and frank look at the Cost Analyses it will take to put Linux on the corporate desktop. In Part I of Corporate Desktop Linux - The Hard Truth he begins with one of the most common misconceptions... that a business can buy a computer without Windows and save money in the transaction."
This guy is trying to make a fair TCO analysis of Linux Vs. WIndows and part 1 doesn't look good for Linux. Now it is time for slashdotters to say that businesses should Install all their systems from scratch or Buy the Walmart systems.
If something is so important that you feel the need to post it on the internet... It probably isn't that important.
When you buy into the Microsoft platform, you are buying endless upgrades for years on end.
When a user bought Windows 3.1, they also unwittingly bought Windows 98, Windows 98SE, Windows ME, and Windows XP. This is planned obsolescense for no other reason except to keep Micorsoft shareholders happy.
With Linux, you avoid that ridiculous problem.
---Technology will liberate us if it doesn't enslave us first.
Too bad this kind of analysis didn't make it into the anti-trust cases....
ok.. so linux is not cheaper because windows is "free" as well.
i think he should take a good look at his support contracts and then figure out just what's wrong with his reasoning here. that's right, his reasoning would be ok IF he was arguing about home desktopts - but he isn't, so what does the initial ten or twenty bucks mean?
of course, maybe the final chapter will be "linux just can't compete.. because linux can't give me huge discounts if i would have said that ms sucks".
world was created 5 seconds before this post as it is.
IDIOT!
He's talking about corporate purchases; where the company buys off the shelf, simply re-buying "their standard configuration" each time another cubicle needs filling. These kinds of customers don't build their PC from bits, you fool!
And they DO worry about tiny differences in price; because they get multiplied out by the hundreds of boxes getting bought by the whole company.
Get a brain before using your keyboard, FFS.
And you seem to be ignoring the fact that no corporation in existence is going to start building their own computers from components. The added labour costs make this the least cost effective alternative.
Why doesn't Slashdot ever get slashdotted?
... after the whole series will be that Linux does not save money for an average company. This first part gives a major point why a company won't save money. Another major point, I think, will be the time spent in helping users using Linux. The average company will have employees who are used to Windows and they will thus need help to get on track using Linux regulary. This time will probably cost a company much more than the licenses for the applications needed if they ran Windows. Another point I see coming up is the time for installing software. I'm a very experienced Linux user, and a less so Windows user, but I'm finding it quite alot easier to install software on a Windows system. And then I'm not talking simple software such as firefox, I'm talking complicated applications such as j2ee application servers (for sandbox development), advanced IDEs and so on.
:).
My two cents
The title of the article is Corporate Desktop Linux. Corporations don't piece their systems together.
So he purposely misstates any facts and says that windows is cheaper because some pricey manufacturers choose to sell windows for less than linux installed machines. This is silly too because you can just buy the chjeaper windows default install and put linux on it with a burned CD you downloaded for no cost. And then throw away the Windows CD!
No, he's stating the facts correctly. Theoretically, buying an N-series from Dell should be cheaper, because they are not including a Windows license in that price. But it's not.
If you want it without a particular option, the price should be less than an identical system that includes that option.
Just like buying a car without a radio. It should cost less than an identical car with a radio. Not more.
OK you are IT Manager of a large company. You have 1000 systems to install. What are you going to do. Buy from Dell/HP/IBM who you know can deliver, Try to organize all the parts you will need bargin shop for all of them Then have your IT Staff build 1000 boxen, Or call Mr. Noname running from the basement and ask him to build 1000boxes for you.
I would probably go with Dell/HP/IBM because it is actually a better value because your time and the staffs time costs money too and you also need to save your butt from management if something goes wrong.
If something is so important that you feel the need to post it on the internet... It probably isn't that important.
We use all the latest technology, XML, ODBC, etc. We still have many Unix based servers.
But the point of the article, and my post, is about linux on the desktop and how difficult that could be to implement (in a large company). I was specifically talking about how most vendor products used in my bank (or any large bank) don't include desktop components that run on Linux. It's different when IBM comes out and says "Were moving to linux!" since they are a technology company. They have the knowledge and resources to make that shift. A bank has NO interest in the progress of technology. They want to use off-the-shelf technology to meet their business needs. Most large non-technology businesses are like that.
$7.95/mo, 200 GB disk, 2TBxfer, MySQL, PHP, RoR.
-cost of Anti-Virus software (that slows the system down)
-cost of anti-spyware solutions. Typically you need two or more cleaners to get the most common ones.
-cost of downtime. Typical desktop PC in a business is down for most of a day many times a year.
-cost of the forced upgrade cycle.
-On top of that, Windows comes with NOTHING bundled. Everything costs extra. Just managing the licenses in a corporate environment is pain!
Add to this the much bigger probability of data loss and theft, and the Windoze solution does not seem like a solution at all.
Bandaid over duct tape. Legacy crap is what keeps people using Win32, there are no other sane reasons.
'Once scientists, even the dim-witted social scientists, get muzzled, the Western Civilization is finished.' - oldhack
That was just part 1, and didn't conclude what you seem to think it did. But hey, don't let that stop your inane ranting.
I would agree that most industries don't look to change platforms or migrate software packages. All it would take is for the demand to exist for the supply of strongly-supported apps to be developed.
Another thought is that with web-enablement of apps, platform independence is that much closer to being an option. And web apps have *definitely* come a long way to providing much of the functionality we have had for some time in regular apps.
Give it a few years, and those web apps will make desktop platform indpendence a reality. You can still program however you choose, and your servers can be whatever you need them to be.
Now accepting PayPal donations!
Unfortunately, the article is very true.
It can be difficult to get pre-installed Linux desktop.
Servers, though, a totally different matter. Here you can make really large savings. Especially when you consider that you don't need all those CALs. Compare a Windows Server 2003 running Windows Terminal Server and having 20 Windows XP desktops connecting to it, to a completely Linux Desktop OS and Linux Server OS solution, and you're biggest saving is in the server area. Heck, according to this article the Linux Server / Windows Desktop would be the cheapest solution!
Actually, I think it's you who's missing the point. First off, to get it out of the way, I hate MS, I use Linux at home, blah blah blah.
Anyway... have you ever worked for a large company, say, a bank or large corporate office, with procurement policies? I think you haven't, otherwise you would know that just telling them "Just buy a couple hundred motherboards, HDD's, CPU's, RAM sticks, cases, monitors, keyboards and mice and have your techie guy working in the basement put them together for you over your lunch break and install Linux on them." just ain't gonna fly. They are large companies, they do business, their business is not computers, but they need computers to run their business, so they look for other large companies that assure them that they are getting solid computers that will get the job done. They're going to buy Dell, or HP, or IBM. They are *not* going to show up at Bob's Discount Linux Shop and order a couple hundred desktops. And they are not going to give their one IT guy back in the server room a pile of components. They are going to go with a large supplier who will deliver a bunch of pre-built, pr-econfigured machines that they can plug into their network, put their username and password in, and get to work.
As for mom and dad and grandma, you try telling them to buy the components and build it themselves. Or telling them to go to Bob's Discount Linux Shop when they can get the same computer with an OS they allready know, and often for a couple hundred dollars less thanks to the discounts the big companies offer that small shops just can't match. They want a computer thay can buy, plug in, and start sharing pictures. They don't know, or care, about Linux or wether it's better/cheaper/sexier. They didn't buy a windows machine. They bought a Dell.
The point the guy makes in the article is completely valid: Unless and until large suppliers like Dell/HP/IBM make computers preconfigured with some flavour of Linux available, and make them cheaper than a comparable Windows box, then Linux will never be 'cheaper' or 'free' to the 99% of people out there who aren't geeks like us.
As a bit of background on me, I also work with Windows 2000-2003 *and* Linux servers for a living, in an environment where we have all our outward-facing machines running Linux and acting as webservers/webapp servers/firewalls/VPN server, and inside the network itself we've got several Windows 2000 and 2003 servers running Active Directory, Exchange, and several proprietary apps that require a server component running on a Windows NT-variant, and a client component running on a Windows desktop. Point is, I work with both Windows and Linux servers and desktops on a daily basis, I have some idea what I'm talking about.
"Two things are infinite: the universe, and human stupidity. And I'm not sure about the first one." - Albert Einstein
If you buy all Linux systems then you will have to train your employees on linux. Just about everybody who knows how to turn a computer on knows Windows. Not to mention getting administrators with a more rare skillset is usually more expensive. I haven't checked the salaries Linux admins command but I know MCSEs are a dime a dozen. Even if they make the same you'd have to at the very least hire trainers for every single department that will be using linux.
Hey, if you can sell the idea to the bean counters more power to you but I don't think Linux will be cost effective for enterprise any time soon.
In most organizations, the most expensive aspect of a F/OSS migration is resistance to change:
WHAAAAAT?!! YOU'RE TAKING AWAY MY POWERPOINT?!!
People grow up with these programs. They devote time and personal resources becoming proficient with them. They don't want that background to become obviated. They don't want to start over. We who work in technology are just the opposite by our very nature. We like change. We like the challenge and adventure of learning new (and better) things. That nature is one of the things that drove is into a technical field.
I personally think the only practical migration is to first migrate to F/OSS apps on Windows, gradually. Then, migrate all those apps to Linux. So that, to the user, Linux is just another application migration.
Research shows that 67% of those who use the term "research shows", are just making shit up.
Any IT worker with half a brain knows that you can deal with your primary reseller. I can get really good HP business desktops sans Windows XP or a Windows XP license for about $500 (dx2000's fully loaded, $400 not). That should be the starting point of pricing for the desktop itself. This guy is spending too much money any way you look at it. Has he never heard of a reseller?
Next you look at the cost of licensing. If you want Microsoft's "Assurance", or whatever they call it these days (in which you can upgrade without fee the next time around), a company my size would have to spend about $300,000. The other option is buy each individual XP license at $176 a pop. Neither of these options include any kind of support. Going with Linux, lets say Novell's Desktop Linux (Suse 9.2 Pro with the LDAP client preinstalled and a shiny Gnome configuration), I'm looking at $80 a license. This includes a little bit of support, and an active community on Novell's official forums.
Anyways, from here you have to figure out how to get around the Microsoft Office lock-in, and decide whether you want to go with Citrix or Codeweavers. But that's an entirely different discussion.
RTWFA - Read the Whole Fucking Article
It looks to me, however, like the Microsoft monopoly has such a stranglehold on the tier 1 manufacturers that it is now not possible for a corporate shopper to save money by avoiding Windows unless they are prepared to go outside the first tier (which brings another whole set of buying issues in organizations with strong procurement rules), or unless the size of the deal is large enough to merit special treatment.
Forget that the upfront cost of Windows or Linux. Focusing on that is like looking and one pebble and using it to describe all the rocks in world.
If you want to look at the cost of any software, Linux or others, in the corporate envrionment you have to considered three main points; managed support costs, unmananged support costs, investment costs. Before you get your undies in a bunch, yes there are other costs but if you slice up all the costs these three can average nearly 75% of the pie.
Managed Support Costs are the things budgeted for. Help Desk technicians, maintenance contracts and such. The cost of a Windows based support person is less than a Linux based support person. There are more Windows people (greater supply). Umanaged Support Costs are those things that are costs generated by unplanned/budgeted items. For example, I sit at my desk and monkey around with getting something to work for an hour. That is a cost to the company both from a resource time spent perspective and from a loss of productivity perspective. THESE COSTS ARE HUGE!
Finally there are investment costs which are things the business will have to do to switch. Imagine a business with a 10 person support desk that is primarily Windows. There are documents, procedures, how-tos, along with the expertise the individuals have built up. Now you come in and say switch to Linux. The business will have to throw away a lot of that list and then spend even more money re-building it for Linux.
Even if Windows and Linux cost exactly the same to support/manage it is a tough sell to a CIO on the topic of Investment Costs. It will cost to much money to switch and for a period of time 1-2 years it puts the infrastructure at risk (new software, unknown problems, etc).
I'm a huge fan of Linux! I have to CPUs at work one is pure SuSE 9.2 and the other is W2K. Even on the W2K I spend my time in SuSE inside of VMware. I am very anti-MS but this is not about MS vs. Linux, this is about costs and risks to switching.
Q: I am short, useless and provide no value. What am I? A: a sig
Yeah, it's a bit of a joke that the guy is using Dell's Windows discounts (or "bias", as I like to call it) as the entire underpinning of a TCO calculation. Maybe he should start noticing computer shops which aren't Dell.
Karma: It's all a bunch of tree-huggin' hippy crap!
Other people have mentioned support, and I'll add to the list:
Quality Control.
I've had whole labs of PCs bought from white-box vendors, and whole labs bought from Dell etc.
We had a white-box lab of 700mhz slot-A athlon systems. after 6 months of running, we had just about every CPU fan die within a 3 week span. The machines were somewhat unstable, mostly due to poor ram compatability. When it came time to cycle the lab, we ended up having to dumpster about 1/2 of them because of problems. (the rest were distributed among grad students who still had PPro200 systems)
Because of volume, Dell/etc can simply do a better job of quality testing before releasing a box.
>>Then we have things such as Exchange, which at first everyone will swear that they need because it has integrated scheduling functions, despite the fact that most corporations hardly ever use the functionality, except for one or two very annoying people who are quickly ignored by everyone else (if you are one of those people, think of that statement as humor). Here is where the price starts getting steep.
I wonder how you could come to this conclusion, but must assume you haven't been around many medium- to large corporate Windows networks. I have, and let me tell you - - scheduling in Outlook, not only for meetings, but also for resources (like conference rooms) and employee vacations, training, etc., was one of the most heavily used feature sets in Exchange. Many companies and small offices rely on the free/busy scheduling facility so much that all employees are REQUIRED to keep their calendars up to date so that management will know the availablity of any colleague at a glance. And I can assure you that, if it weren't for this functionality in Exchange, OSS mail and calendar software would have penetrated much more deeply into corporate American than it has thus far. I've been able to bring Apache, PHP, and MySQL to my current employer's server room, but they will NEVER break away from Outlook/Exchange in the course of my career, I'm sure.
It's only funny until someone gets hurt. Then, it's hilarious.
I'm afraid your logic is ALSO flawed. If you can put OpenOffice on a Linux box, then the fair comparison is to put OpenOffice on Windows as well, therefore negating the advantage of the $379 per machine savings you cite.
:-)
Also, if you're concerned enough about ongoing costs to use old PII workstations across the board, then you should also take into account the electricity these machines burn when compared to a modern PC with better power saving. Granted, P4 workstations are going to burn more electricity, but where I work I think the average workstation is a PIII 1-1.4Ghz machine... relatively low power consumption. Then there's the heat the machines output which must be combated with air conditioning. OK... so you save a little money on the heating bill...
For a corporation, it also comes down to SUPPORT. Your solution is a one-person vendor solution and therefore would be unacceptable to any corporation with any significant size. Where I work we buy Dell workstations because if they break the company can turn to Dell and ask them WTF. We buy HP servers because we happen to think they're the best, and we can turn to HP and ask WTF. We also have UNIX systems which are IBM running AIX... because we can turn to IBM and ask WTF.
Linux has made some inroads... we have VMware ESX servers in our R&D and prodution environments, mostly used in our "production dev" environment... but workstations? Unlikely:
I did run a project about a year ago where I installed Linux on my laptop to show that we could get all of our work done on a Linux machine without worrying about Windows licenses. I did... I was at least AS efficient as I was before (as were others working on it), but in the end it came down to the ability to turn to Microsoft and ask WTF. Until we can meet or exceed this requirement in the open source community, Microsoft will remain on the desktops.
I applaud this guys review... it's part 1. I'll wait and see where he's going with this, but I do suspect Linux is going to do OK in this comparison...
Don't forget the additional cost of the requisite anti-virus system for every PC, and possibly a commercial anti-spyware solution too. Oh, and since the anti-virus slows the system down by X %, you'll need to buy all systems X% more powerful than you need.
With that said, I don't really see Linux becoming all that big on the desktop. Because most of the office users won't start using it at home, simply because 8/10 users plays with their computers in a very different way of what the more geeky types does.
I know plenty of office workers who only know as much about computers as they need to for work. They only have windows with office because that is what they use at work. If businesses started installing linux, then new employees would most likely use it at home, to help be familiar and compatible with work. There are also those people I know who know only the keystrokes and mouse clicks they need to run 1 application, who call the help desk if a window pops up in the way, and they cant close it.
could all of you be patient? I think this guy knows all of this. But this is 4 parts, and he is only doing what everyone would consider part one. small to medium sized business need to look at everything, but this is always the first step, how much do I need to spend up front to get my systems. I just hope you all read the other 3 parts just to see where he goes with it.
Although the previous poster is currently mod'ed as "Flamebait", the post has a very good point...
If I'm a corporation and I am going to buy, say, 100 desktops, I don't go to the website for the purchase. I call Dell or IBM or HP or Gateway directly and get a sales rep for pricing and options.
After the options and prices are set, I would ask how much does the price drop if no Windows licenses are included. I'd probably have a much better chance of getting what I want at that point.
It's one thing to turn down $500. It's another to turn away $25,000.
- Tony
That may be true at your bank, but not at mine. I work for a large new york investment bank where we have a substantial Linux investment.
We are in the "software business", with one client. Ourselves. We write most of our trading software, in house, from scratch. It runs on Linux serves, with the clients being either Solaris or Windows.
Owning your code when it comes to trading systems is a fairly obvious necessity. First for reliability reasons. Can you imagine having a trading system go down and not be able to fix it yourself ? Relying on some vendor to fix would be insane when you realize how much money is at stake.
Second, we need to build new models quickly in order to trade new markets as we develop them. If you wait for some software vendor to decide what is a commercially viable product, you will be at the tail end after all the other banks have made all the money.
Third, many of our models are proprietary. We think them up, write the math, code it and use them to give us an "edge".
Now of course there are plenty of places that we use software written by a vendor, but the point is that if you don't write and maintain your own code for trading systems, it's because you're pretty much a third tier player and can't afford to.
Other places we happen to be going backwards. For instance we recently switched email to Exchange/Outlook across the Firm. But the rumors from the IT folks suggest that the decison to switch has to do more with "reciprocity" (Microsoft is a big client, pays big fees, and demands that we use their software as much as possible) then with TCO / reliability, etc. I've been told that it took over a year to get the Firm comfortable that the Exchange / Outlook that we run is secure enough, that the code we run has been heavily modified for us and is not the same as the code in the retail or other corporate versions.
I don't see what all the fuss is about. Microsoft is paying the vendor to make computers cheaper with windows. It is just part of the packaging like the styrofoam padding, plastic bags and box. Just throw away windows with the rest of the packaging.
When I was young, I had to rub sticks together to compute.
Okay, you can't buy a PC without Windows, at least not from one of the big-name vendors. So if you wanted to run Linux you would need to wipe the hard disk and reinstall from scratch.
But isn't that what most big companies do anyway? Even if you run Windows, you never want the stock installation that Dell put on there. You reinstall the machine with the corporate standard version of Windows (if your IT people have any clue, this will be fully automated).
So I don't see that inability to get Linux preinstalled is a big deal. The main reason to buy a machine which comes with Linux is as a guarantee that all the components have Linux drivers - but you can check that separately.
-- Ed Avis ed@membled.com
So: Linux TCO sucks because monopoly practices mean you're forced to buy Microsoft anyway.
If that's a correct summary, then he's urging acceptance of what is likely to be a highly volatile status quo. That seems simultaneously fatalistic and a bit silly.
Get your teeth into a small slice: the cake of liberty