Slashdot Mirror
Windows Cheaper When Studied by MSFT Analysts
richdun writes "Here is a study done by an independent research firm which claims that under certain circumstances, it is cheaper to develop applications and enterprise solutions for Windows than for Linux. They cite costs from more education, time developing, etc. Of course, the story is quick to state that the whole study was funded and commissioned by our favorite Redmond, WA based software giant. "
(from the article:)
The study compared applications built to run over the Internet on Microsoft's
That speaks volumes: the study is aiming at J2EE and Sun. Granted these are the "certain circumstances" mentioned but MS is taking direct aim at its diminishing server market share with this. They know the desktop is still pretty much a lock-in for the time being.
Here's a clue: don't trust studies. They are generally paid for by people with agendas.
Trolling is a art,
how come when MS sponsors a comparison, and the results favor their OS and/or software over linux, it's just GOT to be a big conspiracy? Why doesn't the same criticism hold true for the supposedly "unbiased" comparisons that are done by linux-friendly companies like IBM and Red Hat?
Oh wait, i forgot...this is slashdot.
"The study compared applications built to run over the Internet on Microsoft's .NET platform to applications developed with J2EE, a development platform backed by Sun Microsystems Inc. (Nasdaq:SUNW - news) favored by the Linux community. "
.NET to J2EE development and called it a comparison between Windows and Linux?
.NET didn't run on Linux!
So, they compared
"favored by the Linux community"? Last time I checked,
Why do I h8 apple?
.NET works on Linux too with Mono. Why not compare Mono/Linux to .NET/Windows so we wont compare apples and oranges.
Never learn by your mistakes, if you do you may never dare to try again
While I agree with previous posts that Linux is cheaper than Windows in just about every respect, I'm glad that his article touches on something.
With the current development tools available for Windows, as well as all third-party utilities/db drivers/etc, development on Windows goes by much quicker.
I'm not talking about little apps that could be banged out as a perl script in a few minutes, but more robust applications that companies need internally.
However, this is just for the present. If/when more people adopt Linux as desktops then more people will learn how to develop for it and more/better tools will become available. Say what you will about Visual Studio, but the recent incarnations work exceptionally well, and they have a large user base. If we could see 1 or 2 similar development IDE's for linux that are HIGHLY ADOPTED (I know there are some nice ones out there, but the use-rate is still rather low), then things could change.
it states
Despite the difference in costs, however, the Forrester report also noted that "many organizations will adopt Linux instead of Microsoft's alternative" because of the expertise they have built up on the Unix platform, Sun's proprietary operating systems used to run computer server networks.
Unix, Sun's proprietary operating system ??
computer server networks ??
Come on.
$ strings FTP.EXE | grep Copyright
@(#) Copyright (c) 1983 The Regents of the University of California.
The Yahoo article doesn't link to it; without seeing the details I'm hesitant to fully swallow any synopsis.
I don't find it unbelievable that some tasks are less expensive to produce under MS products than under Linux, but unless the report indicates other reasons, I'm inclined to believe the difference is due to the trained user-base.
The article does say they interviewed twelve firms (hardly a statistically significant amount), seven of which use Windows & five of which use *nix. I'd be curious to know the sizes of the firms involved and the level of training of the personnel in each of them.
In other words, my question is: Is all else equal? I suspect not.
Allegedly real newspaper headline from 1998:
Man Struck by Lightning Faces Battery Charge
When analysing the cost differences of Windows and Linux, the main advantage to windows always seems to be that little to no training is required, while on the other hand, Linux requires lots of training, with Expensive Admins. However, In the long term, if many companies and schools started using Linux, these cost would come down, as many more people would have experience and require less training. Also the number of qualified people would increase, making the salaries of qualified Linux admins go down.
Anthropic principle: We see the universe the way it is because if it were different we would not be here to see it.
This study takes into consideration more than just the price of the OS. Things like support, salary for developers etc.
This study is sponsored by Microsoft, so it's probably biased as hell, but a Linux system is never absolutely free for a corporation.
J2EE development is slow and cumbersome and it is no suprise to me that .NET whatever that is beat the pants off of it for development efficiency. But considering that the .NET stuff will not run on anything other than windows it is a dead end road. Now do the same study using php against .NET and the tables get tipped drastically.
Got Code?
You can't beat free!
Sure you can. TCO isn't just a buzzword, time spent making things work is expensive for companies.
E.g., installing Red Hat Linux on a server or a workstation is quick and will only need a small tweaking, while the same is far from truth on Windows, where installing the OS is just the first 5% of the job. This applies to other things... buying something which works can often be a lot cheaper than trying to find an open source project which works.
Ok, let's assume that within this select subset of applications, developing for Windows is 25-28% cheaper than for Linux per application. There's also the underlying operating system to consider. Windows, for whatever reason -- inherent security flaws or merely higher visibility / threat risk, I don't care -- is more likely to be attacked and infected by worms and viruses. As a result, the cost of maintaining applications and the operating system is higher. I'm fairly certain such a cost isn't factored into the study, and I'm fairly certain it's a considerable amount -- trying to keep things secured in my company's 250-user environment is hard enough.
Why is it so hard to believe that developing on Windows could possibly be cheaper than developing on a Linux box? For you, as a developer of an open-souce, I code because I enjoy it, project, you are right. Developing under Linux (or FreeBSD) will always be cheaper than developing under Windows... for you.
However, the corporate world is quite often very different. When you have preasures of deadlines and QA, quite often you don't have the time to work with a somewhat less feature-rich, but free, tool. Some tools just make developing certain kinds of applications more efficient, take VisualBasic for example. There's not much else that can compare to the RAD capabilities of VB. Sure, you might argue that there is no long-term viability for a VB app, but long-term viability isn't always needed. Neither is peak performance. Greater development efficiency directly translates into greater profits. Greater profits may quickly overcome any savings that you may have gained from developing your solution on an open source free OS with free tools.
Take for instance a relatively simple GUI application. Say that it takes two weeks to develop the application under a free toolkit like GTK. Now say that it takes only one week to develop that same application under VB. If we use a $60k developer salary (which is only about half of what it actually costs to employ a developer), then we see that one week of time is worth ~$1154. After one month, the license for VB and Windows has quickly paid for itself.
So, for certain kinds of development, yes, you certainly can beat free.
something clever
If you consider the productivity gained by using an IDE such as Visual Studio .NET 2003.
You have intellesense statement completion, automatic code formatting and highlighting, and intelligent help that will pull full documentation on any statement you are typing with one click.
You also benefit from languages like VB.NET and C#, which are very high level and make it easy to write conceptually clear and error-free code.
It makes it possible to become comfortable with a new object model without having to flip though documentation or constantly search the web. You'd be surprised how fast and simple it is to create relatively sophisticated apps that perform pretty darn well.
Amazing magic tricks
its really not that complicated. you can get a subscription to ms dev tools for around $1000/year per developer. if a developer can gain a weeks more productivity in VS.NET than a competitive suite of products, the productivity pays for the software. if 2 weeks productivity is gained, then VOILA! your .NET solution is now cheaper to develop. /.ers seem to forget that peoples time more often than not costs way more in the long run than actual software licenses.
This article really has nothing to do with Linux. It's about J2EE vs. .NET. No surprise that J2EE is expensive. The best Java developers on Linux use much better tools than J2EE and EJBs. A good open source stack with Struts, Velocity, and Hibernate will beat the stuffing out of straight J2EE for productivity. Of course there is also the fact that lots of web development on Linux is done in much more productive languages, like Perl, Python, and PHP. Amazon and Yahoo (on FreeBSD) do it, so it's probably good enough for your lame little site too.
I know I'll get modded down for saying anything positive about MS but here goes:
.NET is infact an even fresher page, intedned form the start to be an internet enabled approach to the net that embraces many different languages as well, including the java reduc C#.
It's not absolutely damning that MS paid for the study. After all who else would pay. But of course the study would not have been released if it were not positive for MS. Still neither of these makes neccessarily untrue or biased.
And its not unreasonable to believe it could be true. MS does make good development tools. And even just a few days ago there was a slashdot post lamenting the sad state of Java. It really could use a third re-do to streamiline it to a sensible coherent set. After all JAVA or as it was called "OAK" was developed to be an embedded OS for appliances. It of course became much more. And it its underlying fundmentlas and syntax were a much needed re-write of C++ (no damn *pointers and &other crap, garbage collection, better OO tools at a lower level).
MS visula studio series may noit be the best tools but they are good tools and present a multi-language interface to programming.
Finally, it is quite obvious that stupid people find it easier to program in Windows. This is not true on linux. Linux requires knowing a lot of intricate details and knowledge of unstable APIs written by other people and not maintained in a consistent or even perfectly protable format.
installing Red Hat Linux on a server or a workstation is quick and will only need a small tweaking, while the same is far from truth on Windows
Look, I use and like linux like the best slashbotter, but lets be real. Either OS you end up using in a corporate environment is going need a lot of time for tweaking and customizing. I wouldn't let a virgin Windows or Redhat install just rollout to an entire enterprise.
My study, funded by me, has shown that Windows is the most expensive to maintain (patch, patch, reboot, patch, reboot) and use.
:). Amazing that IT isn't running around chasing their tails with these new systems like they do with the Windows boxes. Help desk calls for the operating system causing another stupid error and confusing the end user have dropped off. It's more than obvious that those users without Windows have increased productivity. Amazing.
Thank you Microsoft: for security reasons it was determined here that at NO TIME will _any_ operating system gain 100% penetration to the desktops. The datacenter already runs a mix of Linux, BSD, and Netware...
Since the changes started we've rolled Linux out to some people and OS X to others (OS X is my preferred GUI of choice
OK -
.NET is a kick-ass development environment. Even the older non-.NET edition is a lot better than most dev tools out there. Sure, it's pretty expensive - but say you're paying programmers $40/hour (ignore benefits, etc) - the fact you just spent $1200 on a development environment is no big deal: less than a 40 hour week of paying said programmer. And, I'm willing to bet he'll save a lot more than a week of effort by using a better tool.
First, RTFA. It talks about _developing_ applications. It wouldn't really suprise me if it were cheaper to develop applications on Windows.
Visual Studio
Say what you will about the quality of MS, and how buggy/bloated their software is. It seems to work well enough for a bunch of people out there. Their developer programs are excelent (maybe they need to be to cover up their crappy underpinnings).
In the open source area you might be able to download some open source code, and cobble a system together to do what you want... But I think I remember reading a statistic that said something like 85% of all software written is custom, internal, business software. So you might have a tough time finding something that solves your problem exactly... But since it's open source you can modify it to fit - sometimes; sometimes it's more trouble than it's worth.
With things like Eclipse for Java development in the open source arena the gap should close up in that area too (dev tools) - but don't kid yourself; we've still got a long ways to go.
No, but most of the "costs" that are assigned to Linux presume an existing greater knowledge of Microsoft Windows; it never includes the money spent training these people to use Windows originally nor does it try to compare costs of complete training: how long does it take to get an admin for *nix and Windows (respectively) to go from 0 to 60?
Despite what people say about the "learning curve" of *nix, I believe most of that is due to breaking bad habits they learned in Windows. I picked up Unix exceptionally quickly, in part because I never knew Windows very well. So, if one were to remove the costs of Linux training that they have ignore WRT Windows training, the cost is significantly less. Realistically, the costs should be counted for both, not discounted for both...
Someone asked if I had patched against MSBlast; I said yes, I installed Linux.
So, for certain kinds of development, yes, you certainly can beat free.
With that said, have you considered the extremely high cost associated with battling all the worms and virii that seem to be mostly in the Windows world? They have reimaged my machine three times in the last two years! Don't even get me started with the hideous bugs rampant in every piece of software MS has? UGH!
I just finished school and by the time I left almost all of the computer science classes were using linux (red hat to be specific) all the electrical engineering classes were using hpux and linux (debian --old debian) all the programs we made were designed to run on *nix. That implies the cost of educating people to develop on *nix is dropping dramatically.
There is also a large microsoft group called dev hood. They give away free msft products so there is large support for it. But the actuall classes that require msft are all those with minor programming. Most of them use VB.
Can you ping me now?... Good!
In my experience, initial development of Microsoft-centric solutions is often less expensive than Java or even OSS-based solutions. This is especially true for the one-off or "quick and dirty" stuff so prevalent in Microsoft-centric shops.
However, the initial cost savings are eaten dozens if not hundreds of times over by higher total lifecycle costs.
Basically, for every dollar we spend writing, say, a VB/IIS/MSSQL app, we spend dozens or hundreds of dollars maintaining it and rewriting it.
The comparable development using superior technologies does cost a bit more, mainly because you have to hire better developers - people who, for instance, know how to sort an array, without resorting to writing the rows to an invisible ListBox and then setting its .Sorted property to True (don't laugh - I've seen this seriously suggested multiple times, both on comp.lang.basic.visual.misc, and among people I've worked with).
But well-written Java and/or OSS apps stay written. They don't break over time due to DLL incompatibilities or the obsolescence or poor scalability of the toolset or a decision to move to another platform or the virus/worm/trojan of the week.
If it were up to me we would allow the use of M$ development tools only for the "quick and dirty" stuff - prototypes basically - and those would be absolutely unsupported by IT. Any production system requiring IT support would need to be based exclusively on open/Free protocols, standards, and products.
Nonaggression works!
What is the cost when your developers are interupted by worm and virus outbreaks? Is that considered in the study?
Telecommuting! What about socialization?
Its a simple formula.
1) Find the categories your predetermined winner has advantages.
2) Weight those areas heavily
3) De-emphasize or omit any areas the predetermined loser has strong advantages
4) ???
5) Profit!
Sorry,
4) Release study!
This has the added advantage of creating contreversy, which:
1) Increases visibility
2) ???
3) Profit
You are in a maze of twisted little posts, all alike.
When Microsoft funds a study that makes their platform look cheaper, we all laugh and write it off as "bogus". Even the editors are jeering at the results.
Yet when Apple funds a study that makes their G5 look really fast, Slashdot cheers. The savior is here! The headline isn't "Apple Claims G5 is World's Fastest PC", it's "G5 Is World's Fastest PC".
Sheesh. Steve Jobs Reality Distortion Field indeed.
dude, this just doesn't make sense:
.NET Ent 2003: $4657.86
.NET & Windows server ACLs and Windows Clients, we come up with this:
"Take for instance a relatively simple GUI application. Say that it takes two weeks to develop the application under a free toolkit like GTK. Now say that it takes only one week to develop that same application under VB. If we use a $60k developer salary (which is only about half of what it actually costs to employ a developer), then we see that one week of time is worth approx. $1154. After one month, the license for VB and Windows has quickly paid for itself."
Let's really look at company A (the GNULinux company) in a real world example, because like always, when people cite Windowze development-office stuff, they only present part of the picture.
First off, let's use a office with 50 people, of which there are 2 full time developers and 3 full time IS people. The below are current prices from MicroWarehouse at their non-discounted prices.
Now, for office B, first, the Windoze office (I'll leave hardware out of this cuz, especially with Exchange, that's a whole other cost issue):
* 50 copies of XP Pro: $21,747.50
* 1 Win2k file and print server with 50 CALS: $3018.72
* 1 Win2k with Exchange Ent. with 50 CALS: $12,107.44
* 2 copies of VS
* 50 copies of Office XP Pro: $21,747.50
This doesn't assume any consulting fees, whatever, since we'll assume the 3 IS guys are Windoze experts and know how to set up everything.
Total just to get office B up and running so that said developers can develop code: approx. $63,279.02
Now, let's look at office A, the GNULinux office. For basic comparisons, we'll use Redhat 9 deployed.
* 50 copies of RH 9: $0 (cost of 3 cd-r's: approx. $1.50)
* Samba for file and print for all 50 users: $0
* Kroupware (I know, not the best Exchange equiv, but the only free thing really avail strongly for GNU/Linux that delivers scheduling that works) for 50 users: $0
* 2 copies of development environment for GTK-Java-whatever: $0
* Ximian XD2 with Ximian Open Office: $0
Cost (not including hardware or labor) to deploy office B to make snazzy app: $1.50
Cost diff between office A & B: B: $63277.52
Now, plugging in your formula of $1154 for two weeks of development which supposedly pay for VS
Office A has no deficit, since their software didn't cost them anything. In fact, they probably had one less sysadmin, so they might have already saved $60k.
Office B has a deficit of $62,125.02. Their developer, to make them money to cover said expenses will need to work an additional: 2118.71 hours.
Obviously, company B should call M$ or MicroWarehouse and try to get deals on this stuff, but regardless, they're still going to need more people, more licenses, which translate into more costs to even start a system where 2 developers can program something.
So, I call bullshit on this entire forumla and the like because they only take one fraction of the puzzle in to play. This is typical with most Windows vs. GNULinux comparisons. I've seen it happen in the real world over and over.
On Windows, yes, an IDE is a crucial, critical tool. I've written code for Win32 in C, with no IDE (in fact I'm doing it now), and it's a nightmare. The same is not true on Linux. You don't have insanities like pixel-based layouts to deal with, you can actually construct your GUIs in code if you so wish. Or, you can use Glade, which spits out XML files you can load at runtime. This is far, far easier than Windows.
The equivalent then to VB is something like Python with Glade/GTK, and emacs. Or it is to my mind, at any rate. These tools are not integrated into one super environment, but they don't lose anything for it in practice.
The other reason you tend to need IDEs on Windows is because Windows is not designed to be text-editor friendly. For things like COM, it's often necessary to use wizards that spit out huge amounts of auto generated code. There are no such technologies on Linux, as far as I'm aware (in wide usage).
Basically, I don't find it any harder to write software on Linux than on Windows, despite the lack of an IDE. I mentioned loss of context sensitive symbol completion - yes, that's a shame, but OTOH when in a Windows IDE I miss a proper command line, a strong and powerful text editor like emacs, and sane tools and toolkits like Glade and GTK, so it works both ways.
Oh, finally, I found I much prefer the "just get on with it" approach of emacs to the one taken by IDEs, which tend to clutter your workspace with things you don't need, like class explorers, project trees, widget palettes and so on - I like having most of my screen taken up by the text editor, as programming is mostly about editing text.
Anyway, just my 2 pence.
So well-paid developers tend to choose Linux. A well-paid developer tends to be an experienced developer. Maybe through experience they have learned to develop and deploy on Linux. Note that this study says nothing at all about the quality of the software developed on either platform.
Lasers Controlled Games!
I'm sure some day you'll figure this out, but bulk licenses are available to purchase in multiples of ten, I believe. Each license in a bulk purchase ends up being closer to $20 (for the OS) and 15 (for Office). Please, for the sake of the CHILDREN, don't speak of what you know not.
The analogy really doesn't fit, since we're talking about Software Development. Developers should know basic operating system principles and have some knowledge of how to use Unix(Unless the college they went to was brain-dead). *nix systems aren't anything dramatically different either(Like laser controlled what-not), the details just aren't hidden as much. Granted GUI on *nix systems is frustrating since GTK on one system won't always work on another. But you can either do Java GUI, or just make a Perl, etc. front-end. Not a big deal...not nearly half the trouble as many of the MS advocates are making it out to be. And for individual users who use the software that was developed there doesn't have to be a learning curb at all. KDE is almost exactly like Windows, just point and click.
You can certainly compare PHP/MySQP and .Net (and J2EE), as they're all technologies that can be used to implement web applications. I'd agree that they are significantly different (.Net is immature and only runs on NT, J2EE is very complex and runs everywhere, PHP/MySQL is simple), but you can certainly compare them in terms of the overall cost of the project. You could argue that one or the other is a more appropriate approach for a particular application (I don't envy anyone trying to implement robust transaction processing in PHP/MySQL, for example) but that doesn't mean that they can't be compared...
Enable 3D printed prosthetics!