Slashdot Mirror
Why Open Source Software/Free Software?
dwheeler writes: "I've just posted a major update of my paper,
``Why Open Source Software / Free Software (OSS/FS)? Look at the Numbers!''
Many sites give qualitative reasons for using OSS/FS, but this paper
emphasizes quantitative measures (such as experiments and market studies) on why using OSS/FS products is, in a number of circumstances, a reasonable or even superior approach. The paper covers market share, reliability, performance, scaleability, security, and total cost of ownership." Bookmark this for the next time your boss asks.
#2: Reliability.It's more reliable than the crap we sold you last year.
#3: Mobility.You can take this crap on the road.
#4: Manageability.With self healing apps this crap can make itself better, though we're still not sure why it gets sick in the first place.
#5: Performance.It's faster than the crap we've sold you before, or at least that's what we paid ZDNet to say. (w2k faster in 64 meg of ram,LOL)
#6: Security. At Credit Suisse First Boston, all your money are belong to haxors.
#7: Internet. "The Internet is just a fad" - billg
#8: Usability. We've almost caught up with the Mac.
#9: Data Access. Whether it is on C:, D:, or E: we give you easy to remember names for all your hard drives.
#10: Hardware Access. We've finally gotten around to supporting the same hardware that Apple has for two years.
Excuse me, but have you heard about sales tactics? this is just a tactic to move some negotiations between MS and Ford. Do you really think that Ford will switch to Linux on a snap of a finger? what about all their software they're using? servers? clients? what about all their internal applications? you're talking here about dozens of software packages that each department uses their own packages (with the expection of MS Office and stuff like that)
So what this article that you point to is just pointed to MS to say "guys, if you won't lower your ACL copy and Windows XP copy for each machine, then we can invest in alternatives" - but moving to Linux right now (as much as I love Linux and use it daily) will be a totally irresponsible thing to do - specially that you don't have the applications neede, support contracts, or anyone that will walk company size of ford to switch from Windows to Linux, both clients/workstations and servers.
Hetz (Heunique)
You might not like Stallman or his methods, but there is little question as to the underlying philosophical motivation as to what he is doing.
--
Was abolition of slavery bad because it devalued the economic worth of the people they worked for? Is it bad to build a better mouse-trap since you might put existing mouse-trap manufacturers out of business?
Believe me, your protectionist attitude has nothing to do with capitalism, and much to do with the left-wing socialism that I suspect you think you disagree with.
--
--
I love point #1. In .edu domains back in 1999 there was a bunch of Linux machines being used. Like .edu or 1999 results really matter! :)
:)
:)
:)
:)
:)
:)
Point #2 is funny... Of course Windows was the #1 server being used and outnumbered Linux 2 to 1.
I especially love point #4... Gartner produced a survey that said Linux is hardly being deployed in data centers... But it's been widely discredited.
How has it been discredited? Well nothing really factual. Nobody really did any similar surveys to prove a counter-point. No instead 300 slobs living in their mothers basement posted to slashdot.org claiming it had to be made up.
But take this in conjunction with point #3. It's funny how readily the guy believes this despite the fact that it has also been widely discredited in exactly the same way as point #4. Well except the guys who discredited it live in apartments rather than momma's basement.
Oh then there is the wonderful Urban Legend of the Linux world told just after Point #4. I speak, of course, about the Navy ship failure. That's actually been discredited by 300 geeks who own their own homes rather than live in momma's basement.
Then he goes on to compare IIS against Tux in terms of Performance after he just got done comparing IIS to Apache in terms of reliability. I love this! It is one of the fatal flaws of a similar paper written by Kirch and archived at unix-v-nt.org. Only he compares reliability of commercial Unix installations and then immediately extends this to claim Linux is equally as fast and reliable.
Oh in point #2 of the performance area he forgot to mention the "Free" Linux solution cost over twice what the "Forsale" Microsoft solution cost, and had slightly less than double the performance increase. He also didn't mention that the Microsoft software solution was almost a year older than the Linux one.
Heh. Then he falls into Urban Legend #2. Mindcraft was biased... Oh yeah, despite the fact that similar results were obtained by other independent benchmarks that showed serious flaws in the Linux kernel.
The TCO section had me on the floor laughing.
Ohwell, thanks for the laugh!
Yeah, what happens in the .edu domain is important!
:)
:)
That's why Pascal was such a huge success in the business world.
As far as the Gartner group study versus the IDC. You don't seem to understand. The Gartner group study was designed to question and dispute the ridiculous IDC study.
Then you go on. I just have to say... COME ON!? You actually believe that interpretation of the Navy ship story? How a divide by Zero error supposedly caused Windows NT to crash. Not just the server, but all of the terminals, etc.
That story was beaten to death, and it was fairly easy to conclude that the problem was with the application software, and not the OS itself.
As far as performance. You obviously don't read TPC benchmarks but a large part of the solution is hardware dependent. The Linux results were done on an SGI with 16 processors. The top Microsoft results were done on a Compaq with 8 processors. Also in this time frame Windows 2000 Data Center edition has come out which supports 16 processors. In the top ranked TPC-W results we have just such a machine from Unisys running Win2k data center...
Just because you are incapable of comprehending my actual points doesn't mean they don't exist. ROTFL!
This isn't the flip side. This article is comparing Windows 2000 to NT and 9x.
You want a Yugo, you buy a Yugo. I prefer my BMW.
--
Life's a bitch but somebody's gotta do it.
The SGI Linux system is over 3 times the cost, as you pointed out, but it took the Linux solution double the processors (16 in a clustered system, of course) to beat out the single 8-CPU Windows box by a paltry 60%.
Wow, I didn't know Linux was scaling so well to 16 way - thanks for pointing it out. Looks like Microsoft has to chase taillights here.
--
Life's a bitch but somebody's gotta do it.
The SGI Linux system is over 3 times the cost, as you pointed out, but it took the Linux solution double the processors (16 in a clustered system, of course) to beat out the single 8-CPU Windows box by a paltry 60%. Oooo, impressive! Sorry michael, but this guy's essay was embarrassingly simpleminded garbage, and showing it to your boss is likely to get you laughed out of the office. It's already been ripped to shreds by various other posters, so I won't repeat what they've said.
Cheers,
You had some good things to say, but I really have to correct the record on one point that you made, something which was also brought up at the Roundtable:
Actually, the whole motivation behind Richard Stallman's Free Software crusade is pure jealousy and ego. When fellow workers at MIT started taking the research and actually applying it to produce real products to be used by consumers, and of course making money from it, it really honked RMS off. From Steven Levy's Hackers (emphasis mine):
And later, in Forbes (www.forbes.com/forbes/98/0810/6203094a.htm):Not really too surprising from someone who has invested so much energy in attacking anybody who says "Linux" without appending "GNU" to the front of it, is it? How many more hours have been wasted by programmers thinking they were serving some noble goal, when in reality they were just feeding Richard Stallman's childish insecurities and ego? Programmers that Stallman would be thrilled to see homeless and begging for change if they ever even think about making a bigger splash than him. Pity.
Cheers,
The "trouble" here is that there are free alternatives
to everything MS makes. In a couple of years they
will have to compete with Openoffice and Koffice
on price/performance. Do you think they will first
give away Windows and then Office? Because that's
the logical chain if one follows your argument.
As an aside, I think that network-based apps will
not fly because networks have latency and they
go down. Passport already exists and is a luaghing
stock of identification providers (Hotmail security?),
.NET may be good (it may succed on the client but
is unlikely to displace java on the server) and
Hailstorm will be a huge mess. Microsoft's new
business model looks like a money losing disaster
in the making.
Statistics always seem to say exactly what the author wants, so how about this:
A drag race.
That's right, get teams together competing in certain hardware classes and let them compete, MS against RedHat against OpenBSD against Solaris.
Let the teams use every resource they can to tune and optimize and try to kick the other team's butt.
Include competitions for Web Serving, database, crackability, that kind of stuff.
Whenever someone does a benchmark study that shows Microsoft beating Linux, people scream 'unfair' - That the people who set up the Linux box didn't optimize correctly or used a setup that caused the results to be skewed. This would force people to put their money where their mouth is.
I have a programming student who uses Windows 2000. He and I were setting up MySQL on his machine and I was showing him the 'benchmark' feature of SQL, as in
"SELECT BENCHMARK(1000000000,radians(180))". While basically a meaningless test, it did give him some bragging rights and motivate me to optimize a bit when he trounced me and my Linux box, (both with 600MHz and 128MB.)
Who knows, could be fun...
Cheers,
Jim in Tokyo
MMDC Mobile Media
-- My Weblog.
To sum that up as what I meant...
"Free / Open Source does not always mean free or available via FTP"
You miss something obvious when you talk about "Open Source" being bad for single shot, client specific applications.
These already ARE Open Source, most clients pay a service charge to have these application developed and they control the right to use the resulting application (most of those who have any ounce of clue anyway, I'm sure some are stung). In effect, the client is then free to take the work and extend it independently, redeploy it, place the code on the internet, anything. (This would be my experience from development contracts anyway - Perhaps some lucky people are getting away with harsher terms - I suppose Microsoft do).
Those choosing not to further distribute doesn't make it less like Open Source software to them - they can still have someone else work on it later and are not tied to a specific vendor (assuming the documentation is up to scratch). Many just don't realise the benefit of doing open distribution (it tends to save money to have vital, reusable parts of your system available for others to use and improve.
Sorry to feed trolls and all but I thought someone needs to respond.
When someone says the phrase "business model" people often think of various silly ideas that I would put more into the catagory of "scam". No, when I talk about business models I am talking about how you pay for software, not how to "make money" from software.
There was once a time when software was developed by companies in-house. It was secret, proprietory and built custom for that business. There was little forward progress in software methods as everything had to be reimplemented. Eventually, companies caught onto the idea that they could purchase generic tools from "software companies" which were often as good if not better than in-house development. Many companies sprang up to fill this need and the overall quality of software improved.
Open source offers a way to return to in-house development. Companies can have all the advantages of in-house development along with the free testing, debugging and features provided by contributing to community software. Co-operation can replace competition.
To address your concerns (or those who truely have these concerns, as you are obviously just trolling) I can personally state that it would be a relief if my job was not directly tied to the sale of a product. To put it frankly: customers piss me off. In-house development and maintenance is a much lighter atmosphere, without the headaches of release schedules or the lunacy of checkbox marketing requirements.
How we know is more important than what we know.
Mark Twain said "There are three kinds of lies: lies, damned lies and statistics". And I strongly agree, because it seems that for every pro-linux statistic, there's an anti-linux statistic.
Statistics can easily be done badly, or be completely irrelevant for realistic applications, and packaging many statistics together just encourages the reader even more to ignore the details.
--
... Introduction to Open Source and Free Software, a brief (5 page) introductory paper that is understandable to non-technical people. If your boss doesn't "get it", then maybe your boss's boss will.
There's also a book out called Embracing Insanity. It's an outstanding introduction to Open Source and Free Software that management-types and non-technical people can understand. At only 177 pages, it's a fairly quick and easy read.
WE should do opensourced FUD! so everyone can contribute! Lets make a bigger stick together to beat microsoft with!!
BLAH!
Linux measures its power in the number of machines with Microsoft OSes which are reformatted and installed with Linux instead. :-)
Got Rhinos?
- you should know which tool to pick
- you should know how to put tools togheter.
Once you know that, you can assemble a quite reliable and usable desktop (no, not usable as a MS desktop, but more for lack of commercial support than for lack of software/bad software).
I still have the Win98 partition around on my desktop but it is more than one year I did not use it.And when I needed more disk on my laptop, I did not esitated(much) before removing Win95.
The application I use dayly are a lot more reliable than you mentioned (and this includes a rather sperimental mozilla 0.91 and gnome 1.4 without Nautilus ). Up to now, upgrades never left me without a system (I use Debian/testing, but I am careful with what to upgrade).
To use Linux, you shall first accept that an important part of your computer solution is _yourself_. No, this does not mean that you have to be a programmer. It means only that standard linux software[up to now] will not attempt (and badly fail) to think for you. But it will perform faultlessy a miriad of boring and repetitive tasks, leaving you the time to think.
Ciao
----
FB
For people to make price/performance assessments, they should have the possibility to choose between two computers with the same hardware, one running an OSS system, the other running a MS system. And there will still be the issue of the application base (although OSS applications could benefit of their $0 tag).
Servers are a differen matter: Win2000 costs quite a lot. And you don't need so many applications (or games) for it.
As for work-place office desktops, they are a middle ground. MS still has better applications and owns most of the marker, making it difficult for a Linux-based shop to inter-operate with others using MS products. On the other hand, OSS office applications are impoving, and they do not have per-user licence fee (or licence fee at all). And work-place office desktops have less esotic hardware and do not need games, so the lack of commercial support is less an issue, here.
Ciao
----
FB
The counter to your arguement is that even if it is such an application, if it is well coded or contains interesting those ideas can be reused elsewhere. Very few tasks on computers are so specific that no one else will want to do them. There are too many computer users for that. Where I CAN see it is if there is some simulation work or something similar where confidential information must be used when writing the code, such as the physical characteristics and parameters of a new weapon or the details of a next generation supercar. THEN I can see closed source. But any application typically will have appeal somewhere. It's whether or not it's a good idea to let out your ideas that should decide it. Even if the application itself is never used, open source preserves ideas. That is it's real power. Even if in ten years we are all programming in smalltalk or something, we can still look at older code to find a solution to something or an elegant piece of code to base something off of, preventing us from reinventing the wheel. If was a good design, no reason for it not to be used again. If was bad, we can see why it was bad and work to avoid those mistakes. So while open source might not benefit one particular application, it will always benefit the community.
"I object to doing things that computers can do." -- Olin Shivers, lispers.org
The point htis article misses is that it isn't just GNU/Linux vs. Windows.
Commercial vs. Free is, more realistically, AIX, Solaris, etc. vs. GNU/Linux. Even if GNU/Linux can beat Windows in a database test, it can still be whupped (in many cicrumstances) by other commercial Unices.
These are those faulty benchmarks that make one system look better beause of minute differences. Website counts that have 4 decimals, but no approximation of the number of servers on the internet?
I like GNU/Linux, but I'm not going to go advocating it as saying Free Software is better than Commercial Software for reasons other than ethics blindly. I can admit that GNU/Linux isn't right for everything yet. I'm not going to say you should use it for quantitative reasons when most of them are half-hearted MS bashing attempts.
Use GNU/Linux because you want to. Encourage people to use it because it's right. But don't find every study you can, emphasize the ones where GNU/Linux wins and when Microsoft wins (a la servers by IP address) lump GNU/Linux together with commercial UNICES! This is just sad. Linux advocacy for the sake of Linux advocacy hurts everyone.
Lets not be zealots like are closed minded friends. Open source is superior for lots of applications, but not necessarily all applications.
A very specific program for a very small client (only) isn't a good open source project. Open source is wonderful if it works with/for a large communitee. Allow me to explain a little better. Its not a good open source project if it only works/only needed for one company and will never be needed again, because it won't reap the benefits of open source (no one will check the code for errors, or want to expand the code or add new features, etc...).
I am for open/free source, but I just don't want everyone to have a narrow view of open source. This probably isn't the best example (I'm sure that several people that reply to this will point it out). I'm just trying to explain that closed source still has its advantages, however few there may be.
--
"That's one small step for man..." "STOP POKING ME!!!!"
Good quote, too many chars. Seriously, the slashdot 120 char limit sucks!
Why would it be in Ford's best interest to release software??? They made it internally, and aren't distributing it outside their company so why would they push specialized software that they used their own money to develope with???
It seems to me that they would never want internal code to go outside their company, for a couple of reasons.
1) Most likely nobody but other auto makers will even glance at their code (so what's the point then)
2) Other automakers will use their code, and they DON'T have to release any changes if they don't distribute it so Ford would never get any improvements back.
It's worse than just being a weak rebuttal. It hurts his overall argument.
The whole paper is supposed to give 'quantitative' reasons to choose Free Software - i.e. the numbers.
Here he's effectively saying "You don't need to see the numbers, I've already decided for you that they don't matter"
But I also feel that open source has some serious issues that need to be addressed.
1. It seems that quite a bit of open-source software works "just well enough", that is, the nerds can understand and figure it out, but the usability is not there for a technophobe. An example would be the Apache webserver, which a leader in it's class, even when compared to commercial software. However, it's not easy to setup - configuration is through text files, etc.
2. Documentation and Help: It's pretty well known that programmers like to code, not write documentation or a help system. Thus, many open source programs are poorly / not at all documented and have no online help if the user gets stuck.
3. It's hard for open source to compete against pirated commercial software. If everyone had to pay $200 for Windows and $500 for Office, Linux, StarOffice, etc. is an easy choice. However when Windows and Office can be obtained for nothing (and is, by most people), it's harder for open source to compete. If Windows is 'free' and Linux is 'free', most end users will choose Windows.
4. Ease of use and user interface. Open source programmers are not icon designers, color theorists, photographers, or graphic designers, for the most part, and it shows. Open source needs to figure out how to pay people in those fields to improve the GUIs of open source operating systems and applications.
Well, that was kind of long. Hopefully there will be (more) progress in these areas. I know about Mandrake - what is needed is Mandrake times 100.
My biggest problem with your article was use of words like "easier" and "faster." I saw no definitive words, no direct statements. Just statements of superiority over previous MS products. And some hints of superiority over non-MS products.
... sounds like bug fixes to me. I know guys who write patches the Linux kernel and send them in, and I have a funny feeling that the Linux kernel gets publicly released "modifications to the OS core to prevent crashes" a lot more often than any brand of Windows.
... hmm. A few things available for Linux that (more than?) equal the playing field: pgp, gpg, ssl-telnet, ssh, scp, IP tunneling over ssl, ssh, ssh over ssl, Netscape 128-bit encryption (and I'm sure there are more that I don't know about)
Having exactly 10 reasons gave cause for suspicion. And having only a few examples of corporations involved gave the suggestion that your data pool was awfully small.
"Modifications to the OS core to prevent crashes"
"comprehensive security"
IntelliMirror? Sounds like home directories over NFS, if you ask me.
And it really does sound as if you copied it off MS's website...
What's this Submit thingy do?
Not so sure. FUD usually relates to campaigns geared to unreasonably scare purchasers away from a product (IBM vs. Amdahl as the seminal example). FUD is based on distorting facts (See Microsoft's implications that Netware 5 didn't come with RAID support when it clearly did and many many more examples I could pull from their web site if I wanted). I see no deliberate fact distortion in this paper nor do I see an intent to scare people away from NT or UNIX. Rather I see a simple and well written article evangelizing the virtues of open source software.
This article is not FUD. It is good marketing.
LedgerSMB: Open source Accounting/ERP
Then evidently you have a very different idea of what FUD is than I do. But then, I am a historian and a programmer... What do I know?
Arguing which semantic entity is correct is hardly FUD. Misguided, perhaps. But definitely not FUD. If you want to call this article FUD, you will need to do better.
Again, FUD is (by definition!) based on scare tactics. Where in this article does he use scar tactics?
LedgerSMB: Open source Accounting/ERP
GIMP chokes badly on the kinds of large files that are common in design, while Photoshop is optimized for large file sizes. In addition, GIMP doesn't do CMYK separation and isn't integrated with useful color management packages. The performance and feature gaps produce a TCO equation that is decidedly against the "free" GIMP.
As for UI, Photoshop's is generally acknowledged to be superior, which is a significant TCO factor.
With respect to GCC vs. CodeWarrior, I was referring to compile speed, which is a fundamental determinant of programmer productivity. GCC doesn't even have precompiled headers. I was not referring to program optimization.
Tim
Reliability used an oversimplistic methodology, probing only for crashes and freezes based on random character input. This is not a metric that has anything to do with the average time between software failures and the seriousness of those failures in real-world software usage; it's yet another uptime-based quality claim. No commercial vendor would ship software that had been QA'd only through this ridiculously simplistic process, which would miss virtually all bugs.
Performance discussion dealt only with the speed of the base OS platform, not of applications.
The first numbers were based on abstract benchmarks rather than on comparison of real-world software packages. Instead, let's compare building a project with GCC vs. CodeWarrior, or browsing the web with Mozilla on Linux vs. Explorer on Windows or Mac.
When claiming a win on database performance, the article fails to note that the winner, while running on Linux, was DB2, a proprietary product from IBM, not an open source or free software database. Let's try MySQL under load against a commercial package instead.
The third performance test cited was for custom-built software, not applications which are used in the field. Again, it's quite possible the base kernel is faster, but in real-world conditions application performance usually predominates.
The web server benchmarks appear to be for static pages. Apache is known to be slower than IIS for dynamic content.
Security I'll grant is much better on Linux than any flavor of Windows, though a desktop Mac OS (not X) system is more secure than either.
The total cost of ownership issues associated with inferior user interfaces and typically inferior application software performance were not addressed. For instance, compare a shop of graphic designers using GIMP on Linux with one using Photoshop on Mac or Windows, and you'll arrive at a very different TCO conclusion. Ditto for a software engineering team using GCC vs. one using Codewarrior.
In short, it seemed to me a very partisan piece that ignored most of the issues associated with real-world desktop usage.
Tim
Stockholders these days care only about one thing: revenue growth. If a company doesn't increase their profits from year to year (or even quarter to quarter), then the stock price drops and shareholders do silly things like fire the CEO. All very well and good, but Microsoft is in the strange position of already selling near 100% of the operating systems sold with desktop machines these days. With over 90% market share (desktop machines, not servers), the only way Microsoft can grow their business is to squeeze even more money out of their existing customers. This is why they're doing things like "subscription" software licenses that actually reduce value and cost more in the long run.
That's why Linux is such a fantastic competitor for Windows. Linux can't be bought. Linux can't be bribed. Linux can't be sued. Linux can't be undersold. Linux will never run out of money. Plus, as Microsoft is forced to make Windows more expensive (gotta keep those shareholders happy!), Linux will look more and more attractive to purchasing managers.
Actually, giving Windows away would probably be their best move in the long run. The shortsighted nature of their business makes it impossible, though.
This
Historically, proprietary vendors eventually lose to vendors selling products available from multiple sources, even when their proprietary technology is (at the moment) better. Sony's Betamax format lost to VHS in the videotape market, and IBM's microchannel architecture lost to ISA in the PC architecture market, because customers prefer the reduced risk (and eventually reduced costs) of non-proprietary products.
IMHO, price is the reason that Open Source is kicking ass. Betamax, Microchannel, the Amiga, and a hundred others lost to inferior competitors that were simply less expensive but good enough. Consumers are almost universally concerned with getting the best "bang for the buck", and nothing delivers that better than Free Software(tm).
That's why Microsoft is so paranoid about Linux and the GPL. There's absolutely no way that they can compete with a superior product that's free (as in beer). People only buy Microsoft stuff because it's perceived as having value. As Windows licenses get more and more expensive, that value proposition gets shakier and something like Linux that's free (as in beer) looks a heck of a lot more attractive. The fact that Linux is faster, more reliable, and more scalable is just sweetener that helps seal the deal.
Thanks to Mr. Wheeler for this beautiful progress report. It's news like this that keep us Open Source advocates going.
This
Sorry, I am too busy at the moment to verify his references. Some of these don't look credible anyway.
(Two weeks later he makes an order for some MS products based off an advertisement from MS, without thinking twice).
Something that should probably be addressed... It says "There are many other factors, but I've found no evidence that they usually cancel the factors listed here." Well, obviously Microsoft and Sun have in their TCO articles, so if you want to rebutt them, you're going to have to find either (a) evidence that they don't cancel the "price of software" factors or (b) show the evidence that Sun/Microsoft present to be flawed. Saying "I've found no evidence" proves nothing.
--
Convictions are more dangerous enemies of truth than lies.
Convictions are more dangerous enemies of truth than lies.
- Nietzsche
Sure, open source works in practice, but will it work in theory?
So the next time somebody says microsoft is taking over the world...
Looking over the "paper" I noted some interesting things on just a quick viewing: 1)under "Performance Date" item 2 "GNU/Linux was the May 2001 performance leader in the TPC-H decision support (database) benchmark (``100Gb'' category)" .edu domain were used..." and if you look at the report there are NO .com domains represented. Well, gee. I wonder why there are so many linux boxes in the report?
Just pointing out that statistics represent those that present them.
Um yes, they did, but they did it on a machine that costs $948966.00. System description It was one of the most expensive machines in the running. The number 2 machine is an Win2k / SQL Server 2000 machine for a third the price. The Top Ten price / performance list is dominated by Windows 2000 / SQL Server 2000. TPC.org
2)The count of web servers in operation is a bit misleading as the source of the information states that "...host addresses of the
There are very few real things in this world...this isn't one of them.