Slashdot Mirror
Win2k Cheaper than Linux
An anonymous reader writes "According to this story, Win2k costs an average of 11%-22% total cost of enterprise. The study showed that the initial investment takes up less than 5% of the total cost. Linux did beat Win2k in one category, Web-serving." Man did this thing get submitted a lot.
Linux costs me anywhere between 1 hour and 5 hours to download an iso of my favorite distro. Win2k costs me 5 minutes to burn a CD-R and 30 cents to buy the blank disc. Overall I would say that since with a minimum wage job I can make 6 dollars in an hour that win2k is by far the better value.
The GeekNights podcast is going strong. Listen!
I can define TCO my own way, but it might prove that BeOS was king (yeah, right); and other's may define it their own way. We'd need to know exactly how they defined TCO to know.
if you only buy a single copy and then install it on your entire network!
From the comments under the article ('BSD user'):
Reference: Here we read that Mainstream support for windows 2000 servers will end 31 March 2005 That's only 2 years and 4 months from now. I don't remember seeing a 'use before' date on any linux servers. Do you?
Readers might wish to balance this article with the rest of the story, found here.
IBM thinks differently in this paper and so does CyberSource here.
As a technologist I'm very sceptical to economic calculations. I believe that they can be twisted in any direction.
There is a principle of uncertanty. Of the three items cost, time and product you can only know one. So if you want to know what product you'll end up with, you can't know the price or time...
Anyway, it is good to point out that Linux systems has problems in the management area. But still, people are working on it.
keep in mind, we had the full $2k/year MSDN subscription for each developer, paid each year, as well as some very experienced staff on hand... MS charged us $150/h to talk to us about a problem that we were pointing out in their CMutex MFC class (a bug they later admitted to) This was back in 1995 or so before MS jumped on the newsgroup bandwagon. At any rate, i wonder if these kinds of fees factored into the TCO?
I love this comment on article on CRN's website:
"It just sounds strange that this article claims a five years study using Windows 2000. As of today, this study should have began by Dec. 1997 ! That means getting Windows 2000 two years in advance. "
So they must using a SWAG (Scientific Wild Ass Guess) to come up with it TCO figures.
I also wonder how rough Windows 2000 was in 1997! Could it be that these figures are made up!?
TWW
"Encyclopedia" is to "Wikipedia" what "Library" is to "Some people at a bus stop"
You get meta-karma, for actually using the word "balance" in the same sentence with a link to the register. I was impressed. If course, it's unbelievably funny, but I was pretty damn impressed at the effort.
On another front, you can get well-balanced news stories here.
Firstly, a server license is more than $150. Second, this is a TCO study. You may not agree with it, but you clearly don't even know what TCO means if you just look at the cost to install and configure the OS on a single server.
:)
Considering all the licenses for W2k where I work cost less than one of our tech support guys' salary (and we have several of those guys) the TCO mostly depends on incidental costs from running linux or windows (ie/ if windows requires one competent admin at $60k CAD and linux requires twice as many, which has the lower TCO? But then factor in how much time those admin's are required to patch the servers and it may change - as you can see, it's not a simple thing to calculate!)
I'm no expert on TCO (i'm a programmer/analyst, not a CTO) but you know so little you really shouldn't even be posting on this topic. Shut up and read what some real admins have to say and maybe we'll all learn something
Robots are everywhere, and they eat old people's medicine for fuel.
We see a similar effect where I work, an NT box costs us about 30% less to run than a Solaris box.
Why?
There are less mission-critical systems running on NT, so there are less DBAs, less backup, etc. The print server sits in the corner and gets a 3-finger salute if it plays up, so it's cheap to run. The mission-critical boxes, running web servers, databases, etc can't go down, so we have administrators to look after them.
IMNSHO - if we normalized for what each box is doing, Linux and Unix are cheaper to run.
Alan.
Without some details it's impossible to tell either what these results were based on or the specific areas where win2k was found superior to linux. I didn't see a reference to the actual study, so there is no way to gauge the validity of the results. There's just no meat to talk about with this marketing blurb dressed up as a news report.
The story mentions that downtime contributes more than 20% of the TCO of a system. With uptimes of months to years for *nix boxes; whereas you are strongly advised to reboot Windows boxes on a regular basis, where is the logic that 23% of the TCO of a *nix box comes from downtime?
We have linux servers at work that have downtime every 6 months for servicing, and then only for a handful of hours. Other than that, they don't come down at all. I fail to see how less than 1 day downtime/year (planned, at that) can contribute 23% of the TCO of the system.
2 sysadms at ~$70k/yr = $140k/yr. $0 for licensing. That would make downtime cost roughly $32k/day (23% of 140k, assuming 24 hrs downtime/yr). If you house something critical, like your CRM system, on 1 machine, and it goes down, I could see that. Then again, that would be your own damn fault for having 0 backup/redundancy.
There's a lot about that article that doesn't add up, and not just the 5 year study on Win 2000...
Actually, a corporate environment is more likely to stay with an old operating system than an individual or small business. There are still plenty of companies that are still using NT4 with Novell clients, or even Windows 3.11. Hell, there are still many (inventory, purchasing, etc.) systems that run on mainframe-type unix terminals. Agreed, most companies don't go 5 years without upgrading but there are certainly some that do.
Enigma
Wonder why Mac OS X was left out of the quotation.
Sheesh - it was a FIVE YEAR study and Max OS X hasn't been out that long. Oh wait... neither has Windows 2000. In fact, Windows 2000 will not be supported five years after it's release date.
Oh I get it. Windows 2000 doesn't cost anything to support after 5 years, since your forced to upgrade at that time.
I expect a lot of side-taking on this one.
But I cannot see how they can support the argument except that at the moment, there are simply more Windows administrators and techs out there than there are Linux administrators and techs. What's more, I have encountered people who proudly make statements like "Microsoft Only" as if it were some status symbol or major accomplishment and who won't even go NEAR a machine running anything else as if it were diseased and might infect his mind. (Brings to mind certain flavors of Christianity)
But as there are more Microsoft-supporting professionals and so many of them are still out of work, it stands to reason that the TCO is low over 5 years... except one thing-- will Windows2000 still be supported in 5 years or will their license terms change again encouraging [requiring] upgrades to their latest OS? So yes, MS people are more available and will accept lower pay. Linux people are still more rare and generally expect more pay because we know a bit more... and usually know MS in addition to other OS's pretty well.
You still get what you pay for, for the most part. But the TCO figure is a very subjective thing... and has anyone asked if this was also yet another MS supported study?
Well, it did take me about 6 months to learn how to parallel park smoothly. But - once I had learned, it was in fact much easier, because the clutch gives you an added dimension of control as you slip into a tight parking space. I got to the point where I could park the manual in a space 6" (15 cm) longer than the car. No one with an automatic trans could match that.
My experience with Windows products pretty much parallels (ha ha) this: easy to learn. Hard to administer.
sPh
According to the analyst responsible for this study; "Linux requires more care and feeding, basically...".
Read more at InfoWorld.
How does Linux require more care & feeding? I don't understand, my experience has been the exact opposite. Whenever I patch a Linux box it continues to function properly, similiar maintainence on a W2K server (with a subsequent reboot) invariably leaves me with a new problem. BTW, patches to W2K servers are far more frequent and require longer download times than any Linux patches -- even when a new kernel is required Linux is still faster.
I think Giga has the right perspective here, if you don't know what you're doing of _course_ it's going to require more care & feeding. I'm eager to read this report, there is another MS sponsored study coming out 1Q2003 that should be equally interesting....
Eric Sarjeant
eric[@]sarjeant.com
It takes a sysadmin every bit as competent as a good Unix sysadmin to PROPERLY administer a Win2K server and its associated workstations. It's a fallacy that you can hire cheap newbies to run a Windows network. Instead of having the problems fixed, scripts written to have certain things get fixed automatically, you get reboot monkeys.
Oolite: Elite-like game. For Mac, Linux and Windows
is that you don't get ANY points for installing it. You get 1 MS Licensing point for each copy of XP, 5 for MS Office and 10 for 2000 Server. No points at all for Linux. How can it be good for your business if you can't get any points! And levels. When you reach certain numbers of points you get new levels.
I think the new MS licensing agreement was actually a RPG system that fell into the wrong hands.
For a good headache...
Part of the lower cost comes from the factor of scale. If you're looking to do some consulting, well Microsoft has a massive and undeniable lead in the number of users- so you start up a business to take advantage of this and offer services for Microsoft software.
But everyone else is doing the same thing, so you have to lower prices and they lower theirs. (This is overall mind you, not pinned down to any two support services) Microsoft products are also quite easy to manage on the whole. Especially since Win2K came rolling in, plus with NT4SP6a you shouldn't have too many major server problems either.
Everywhere you go you can find all sorts of Microsoft camp product support. Once you learn one Microsoft product you are well on your way to knowing another.
Many corporate level packages also come on Microsoft (ERP, etc.) so that gets added into the mix as well - if you want a Linux solution you are really going to have to take the long way around for a lot of this stuff.
Linux is doing quite well, but entry into the Linux world is like running into a brick wall for many. There are far fewer Linux users around and the system is totally different from what most people are used to. There is a staggering amount of things to learn when taking on Linux, kernel recompiles, following the chains of dependancies, all of this takes time to learn and internalize. Most Microsoft type products are a matter of getting the latest service packs.
So there are fewer Linux users and fewer people overall familiar with Linux. The cost of finding someone to help you is going to be higher. Plus, I would argue there is *far* more to learn so you're going to pay the high priced people even more.
This presents a massive total cost barrier for those who would seek to save licensing money by switching to Linux. It is far easier to pay out to a software company for support and pay cheaper mainstream consultants and get things done than it is to start entering this whole new world of OSS. And you'll have to keep paying out more money to expensive consultants and employees to keep up-to-date, even though the initial costs are cheaper.
Then there's all of the little things that Linux can't quite do yet. Incompatibilities with the mainstream software products, pieces of software that just aren't available or which just aren't up to snuff when compared to the MS world. Add these in as indirect costs - even if you get the money to start up with Linux these little niggling issues will make management wonder why they bothered. Finance is not going to be happy without running Excel, the VP is going to be annoyed by not being able to access his IE only stock market site.
On the flip side, if you happen to have employees that known their Linux and know it well, there are definitely benefits to be had. If you want to add a new web server, W2K Adv Server is going to cost you more than the hardware and your Linux-savy employee can probably get an Apache server running nice and easy.
The problem is Linux is just not quite popular enough yet so these gifted people are hard to come by. Trying to insert Linux into a corporate world of Windows raised folk via consultants is going to mean huge dollars - basic stuff that everyone at least sort of knows how to do in Windows may require more consultant hours for instructional purposes.
But, even as the article mentions there are places were Linux is making itself cost effective and useful - like webserving. These tasks should be Linux's thin-end-of-the-wedge. Slowly get Linux in there for these tasks, and then maybe it can take over one more job, then another. Sys Admins can slowly learn more about it and become more experienced. Eventually that TCO is going to balance towards Linux.
There is a long ways to go though - and screaming that all MS users are idiots and they just don't realize how far superior Linux is, is counter-productive. The technical snobbery that often goes on (knee jerk MS bashing, even near-religious fervour found within variations on Linux, newbie bashing, etc.) helps nothing. The rest of the world will just ignore Linux even more and continue on doing their business using MS and closed-source products that they are comfortable with and *that work* as often as not. They *really and truly* don't care what software they use as long as it works, and as long as it is cost-effective to use it. Most business need to use computers, but what computers they use are irrelevant to them. They just need to, well, take care of business.
Find ways that Linux helps them to that in a cost-effective and friendly way and I'm sure more and more business will bite.
Totally worthless... they will say whatever the company paying for the study want to hear.
... one linux admin can handle WAY more linux servers than 1 windows admin.
I'd like to see this study with 10 windows vs. 10 linux servers, or 100 vs. 100
In my experience it's easier to backup and restore Linux based systems. Fair enough with Windows you can backup the registry hives but that's a lot trickier than just copying a few text files. When Windows NT/2k/XP won't boot (BSOD on bootup) you're often up a creek without a paddle. At least with Linux you can get the system up with a bootable CD or boot floppy.
The thing they don't seem to mention is the volume of staff needed to admin the two different platforms. I think that 1 MCSE (or equivalent) per 5-7 Win2k servers is probably a common ratio at many companies (that's about what my company works with), while you can have 1 *NIX admin manage 20+ Linux/BSD machines with ease (my group has 5 people managing 96 FreeBSD servers). Many companies don't understand this, they think that if you just *throw* more staff at the issue (untrained staff at that), you are getting an acceptable TCO. One good *NIX admin hardly needs to admin his boxes at all.
Fully licensed blockchain psychiatrist
Dunno, my experience is a lot different, and I've seen both large Unix and large Windows environments.
Without any fancy tools, administering a large number of Unixy boxes is easy, whereas administering a large number of Windows boxes is hellish.
With fancy tools (which are available for both environments - see Tivoli) you can set things up so that operators can do just about anything as long as nothing breaks. When things go wrong you end up having to revert to the standard admin. tools anyway. Unix is fixable, Windows a nightmare.
It is tempting, if the only tool you have is a hammer, to treat everything as if it were a nail. - Abraham Maslow
You laugh? I worked at a company back in 2000 that put a serious investment into developing a website, with a Windows 2000 back end of course. This was a Fortune 500 company, so of course Microsoft conned them into hiring a bunch of MCSDs to code the thing at some insane price. It ended up costing several million for the whole shebang, which included space at Exodus for a 4 webserver cluster, behind an F5 and dual PIXs, with the database running off two quad Xeon boxen sharing a SAN. This all ran Windows 2000 Advanced Server so they could use the HA clustering functionality.
/.ed I'll be able to take a look =). I'll still take it with a grain of salt, as I would any report comapring the TCO of either of these products. Lies, damn lies, and TCO studies indeed.
Anyhow, I worked a later shift, and got to monitor in the evenings. Every evening, without fail, I watched each and every machine in that HA cluster get rebooted =). "Scheduled Maintainance" I imagine.
Don't get me wrong though, this isn't an MS bash. I'm and MCP, RHCE, etc, and use both Windows 2000 on the server and desktop, as well as Linux on the server and desktop. Each has their place.
Maybe when the server hosting this report isnt getting
It might be worthwhile noting that real studies, which we can look at, unlike this one, and which aren't backed by MS, show that Linux has a lower TCO:
http://www.cyber.com.au/cyber/about/linux_vs_wi
http://www-1.ibm.com/linux/RFG-LinuxTCO-vFINAL-
social sciences can never use experience to verify their statemen
We can see the great benefits of a MS solution firsthand by the performance of your server.
The site www.crn.com is running Microsoft-IIS/5.0 on unknown.
What is the TCO of replacing that smoldering hunk in the corner, guys?
well the original article was about tco of running linux as a server, and not really aimed at developers.
one thing to not about Free software is that the support base from the community is huge. as a result when you are having problems there are many more resources available for you online than there are for proprietary software. also people developing Free software are more likely to admit bugs and problems with their system than those who close their source to the public.
my own personal expirences have shown that developers in the linux community are more likely to respond to you personally than those from say microsoft. take for example a problem i was having with a network card. i was getting strange errors in syslog and i wasnt sure what they ment. i poked around on the net and i couldnt figure out what was wrong. in a last ditch effort i emailed donald becker. perhaps you've heard of him, he writes most of the linux network interface drivers and he came up with a little clustering concept called beowulf.
well i emailed him with the problem i was having, and do you know what he did? he didn't ask me for money, or a credit card number, or a beer. he emailed me source code for a diagnostic program. i emailed the results back. this continued for a couple hours and eventually we determined that the nic was bad. oh did i mention that he responded to my initial query within an hour?
now i ask you, if i emailed support@microsoft.com and asked them for help with my nic do you think the guy who wrote the network card drivers for windows would respond to me personally within an hour to work out my problem for free? this is the difference between support costs in windows and linux. you might not appreciate them, but i do.
-- john
One of the common citations that are bandied about is that Linux admins make more than their Windows counterparts. But, the evidence seems to contradict this "wisdom". Most of the Linux admin jobs that I see posted offer lower salaries than comprable Windows admin positions. Surveys, such as this, also indicate that Linux admins are actually paid less than their MCSE counterparts. This naturally begs the question, are Linux admins truely more expensive than the Windows admins?
Another issue is the "difficulty" of administrering Linux, as compared to Windows. While, there are some valid arguements to support this hypothesis, there are also some important details that are seemingly ignored. That is, the difficulty is in fact due to unfamiliarity. Windows admins are unfamiliar with Linux and it is therefore more difficult for them to administer it. But, were these Windows admins born knowing how to administer Windows? Is Windows truely so simple that they can do it without any prior knowledge?
No! The fact is that the Windows admins have had specific training in administering Windows. They have gone to classes, MCSE Boot Camps, seminars all about how to manage Windows. They also have a bookshelf FULL of Windows administration books that they have studied. Now, after all that, Windows is familiar and relatively easy for them to administer. I challenge anyone who makes the difficulty claim to build a bookshelf of equal size to their Windows one. If these people read just as many books on Linux as they have on Windows Administration, they would not find it any more difficult than Windows. This would likely be true even without any Linux classes or Linux Boot Camps.
It has been proven by a legion of CNEs who find Novell no more difficult, in many cases far easier to manage than Windows. Yet The same Windows admins will say that Netware is MUCH harder to manage than Windows.
Also, on the subject of training etc. These TCO reports always factor in the expense of Linux training. However, they do not seem to factor in the cost of Windows training. Let's not forget that the books and the classes and the MCSE boot camps cost a lot of money. Even if that money has already been spent, it must be factored into the TCO. These MCSEs were not born knowing how to administer Windows 2000. It costed thousands of dollars each to raise this generation of MCSEs. In most cases these training courses were paid for by the companies. How can they be simply ignored by the TCO studies? Are these MCSEs going to live forever, or are they going to be replaced by a new generation that will have to aslo be trained at a cost of thousands per head?
I saw it in IDG.net. It's pretty funny...
Well for nearly 11 years I have been in the fileserver world. I touched lots of file servers. From old ancient LANtastic and Netware 2.15, going through most Novell flavours up to 5.0. For 11 years I worked with, administered, tweaked and crunched so many different file servers that I don't remember all of them. Lots of Novell flavours, OS/2, NFS on Solaris and Linux. I worked also with Windows "solutions", from WfW up to Windows2000 Server. From all these I sincerly prefer Netware. Netware is far better and manageable than any other file server system. Naturally as Novell did it specially for file servers. However there is a problem with Novell. Its prices are prohibitive for many customers. But, if your work highly depends in file server services, surely the TCO is far lower than everyone else.
Among all the systems I used, the most crappy, cumbersome, crash-proned, time consuming and nervestraining was M$ crap. It came up into hanging a whole local network, just because M$ thought it could play at will with TCP/IP stack. But there are tons of stories about the crap. Let's just pick the most recent.
In April this year, I met a medium-sized Compaq server in one highly important organisation. Compaq's dealer sweeted a lot to have that lovely machine there. And sweeted even more to have it working. The thing worked, naturally, on Windows2000 Server. I was asked to tweak the crap so that several problems were gone. And the problems were: workstations loosing connection with the server, Apps frequently hanging up, file transfer working slowly (in a 100mbits network it looked much like 10mbits), and a episodic events with the machine crashing.
After some administration we came up to the conclusion that the machine was going into sure doom. The DNS was crashing every day, WINS and SMB were giving wrong packets into the network, the file system was getting wrong data, user accounts were not freed, CPU never lowered behind 30% and lots of many other problems. Besides we found that, everyday, 30 minutes of workday was lost on backing up data (it was a damn important server) as no one could work while backup was going on.
Well, we created a backup server, curiously on Linux, but with the objective to reinstall Windows2000 on the main server. We lost ONE week trying to do it. As we discovered, the original installer had also huge problems with that machine. The machine was simply unable to work stable with Windows2000.
Considering the pros and cons I decided to use my old weapon The Penguin Dancing Samba, against the huge oposition of many people. However the situation was Hell in Flames and there should be a fast solution. So the bosses agreed the change.
Well I had a whole day of headaches to install it on Compaq's RAID. Also I had lots of trouble creating a secure, stable and automatised environment. In the whole, it took me 2-3 weeks to do all the work.
Today, nearly half-year later, the admin approaches the server 1-2 times in the week. Most work is log checking and some rare tweaks in the configuration (mostly adding users), the machine carries several early warning scripts in case something goes wrong. Backup is completely automatic. With the exception of one single user (some mystic problem), everyone works without hangups, crashes or lost connections. The system lives perfectly in its 100mbps network and the problem of slow connections is forgotten. Besides, the average load of this machine is just 3% and it now carries also a MySQL server that is frequently used and which, in the future, may substitute many file server tasks.
Is this the the higher TCO they talk about?
When you purchase and install a Win2K product, you give Microsoft the right to "audit" you. They performed one of these audits on a company I worked for. I am certain we owned all of the software that was on our machines. We had a corporate policy of no piracy, buy what we need.
We just couldn't produce EULAs for 13 out of over 600 products. Their lawyers also wanted $6000 for the MSDN copies we had. These guys don't seem to even understand Microsoft licensing and appear to be trying to squeeze you for every cent. I had to fax the MSDN user agreement stating that MSDN CDs could be freely distributed within the company. It did not seem to matter to the law firm that we could produce the CD covers for the other products. No EULA, no credit. It cost the company $13,000 to settle. The lawyers got 2/3rds of that for their "work". The remaining third went for purchasing software which I feel we already owned.
I felt scammed and violated. This ticked me off so I looked for alternatives. I discovered FreeBSD. I installed SAMBA and had the same fuctionality as a Windows Server without the risk. I had to buy 2 Samba books to figure it out. I had to reinstall FreeBSD multiple times until I figured out how to do it. I can do it now in my sleep. It is not that FreeBSD is harder, it was just unfamiliar.
If you think this is an isolated incident, it is not. Audits happen everyday. Sometimes, the target really deserves the attention, sometimes it is just Microsoft biting a hand that feeds them. Sometimes, Microsoft's lawyers go over board and put the squeeze on a non profit or school and then people squak at Microsoft. Then there are a number of small companies that, unwittingly, find themselves in a bind.
There are alternatives to some of the Microsoft software. I suggest to everyone that will listen to use the alternatives first.
I find it amazing that all these instant pundits and press-release-repeaters haven't noticed that the IDC study was funded by Microsoft; this could call the results into question.
At least at eWeek, someone noticed this:
"Study Finds Windows Cheaper Than Linux (continued)
"Many drivers of cost need to be uncovered in such an examination and evaluation, and the 'risk/return' trade-offs of Linux versus Windows may not be as obvious as they appear at first glance," they said.
ADVERTISEMENT
The fact that Microsoft paid for the research is likely to be used as a weapon against the findings by some in the Linux community and will also elevate the debate about how valid calculations of total cost of ownership are for any given comparison.
A Microsoft spokesman confirmed to eWEEK that the firm had completely sponsored the White Paper but said that IDC had controlled the methodology, data and findings. IDC analyst Al Gillen agreed, telling eWEEK that the firm undertook a lot of custom research for individual companies and customers."
And Galli also goes into detail about the methodology, so you can have fun picking that apart.
I'm a Linux/Unix admin. And one of the things I tell people when I'm interviewing is that being an admin of these kind of machines isn't that difficult because they are reliable, and when there is a problem, the problem usually has a clear solution (ie, good error messages, much much better error messages than windows). Anyway, as far as TCO goes, since admining these boxes isn't that taxing, I can do _other_ things like program. This is something that I have never heard of a MCSE doing.
Btw, I have a part time admin job (3 days a week) of 63 Linux machines, and the other days of the week I help out with another 100+ machines (admined by one person full time, who also programs), as well as supporting a number of lower priority machines.
If you read the article a bit more closly and check the comments for referance, you'll notive the article says that this study was done over a 5 year period of time. Windows 2000 wasn't out 5 years ago making this rather impossible and thus pretty hard to believe. And I can imagine that starting with Linux 5 years ago and using that till now probably would cost more than it would to start now and carry forward 5 uears because so much progress has been made. Were upgrades allowed? This article is very light on the details. Would a service pack be allowed then? Wouldn't this make Linux better because for free, you get better and better upgrades. Win 2000 only gives you a few services paks, unless you upgrade to XP (ha!)
So in the end I am really confused at how this is even possible and sort of able to believe part of it because of the severe age-ness of it. But really. Come on, we need way more detail before I'll actually believe it.