- Site menu:
- Latest Discussions
- Latest Newswire
- Archives
- Submit Story
- Story HOWTO
- Syndicate
- E-mail Alerts
Free Software in Reality Isn’t Free
[Editor's Note: See Paul Ferris' editorial in response to this.] Linux when first appeared though did manage to make quite a good name for it, however failed to live-up to its reputation in the years to follow. It was to spell the doomsday for commercial UNIXes of the likes of Solaris, AIX, HP-UX and IRIX operating systems. But since its inception in 1991, things have hardly changed in the corporate scenario. Sun Microsystems Inc. is still the UNIX industry leader with around 34% of the UNIX market share (closely followed by IBM in the second-place) with Sun Solaris being the most widely used commercial UNIX. So, what went wrong with Linux? Stable kernels appearing late, innumerable number of useless rewrites of kernel code in a few years, unavailability of proper marketing sources and failure to fulfill the predictions of the community has not only slowed down the number of users in the recent times, but has led to a more serious and deeper thinking about Linux’s future in the recent years to follow. Linux was initially designed to run on the Intel 32-bit platform. Though it has been ported to a larger number of hardware bases later, it still lacks the efficiency of NetBSD when it comes to free portability. But the main trick in using Linux or any other free system-level software lies not in its technical features and advances. It lies in its commercial usage capability. The most important question in the corporate world is “who is to support the systems which run Linux or any other free system-level software?” Take a look at the Wall Street and here's a snapshot. Though industry-leaders (in the Linux arena) Red Hat looks quite steady with stocks rated less than a 6.0 US dollar value, other Linux-distributing firms have less than 1.0 US dollar value. So, what’s wrong with Linux and other free software when it comes to NASDAQ? Why aren’t the corporate still embracing Linux freely and openly as when compared to proprietary hardware and software vendors like IBM, Sun and Microsoft? Linux (also FreeBSD, OpenBSD, NetBSD) lags from commercial software like Microsoft Windows 2000, Microsoft Windows Server 2003, Sun Solaris, IBM-AIX and others when it comes to supporting the installed bases. Linux (and the community as a whole) provides no guarantee of support and maintenance to its users whatsoever. Once installed, who is in charge of proper configuration of these systems? Even once when properly configured, who is there to maintain and support these systems 24/7 around the year? Corporate managers investing their company resources while purchasing type of system-level software, application-level software and the bare hardware ought to think about the type of investment he is choosing to make rather than the often more lucrative amount of investment. Purchasing commercial hardware that runs commercial software is the best choice available. A firm choosing to purchase 5 Sun UltraSPARC stations is better off than purchasing 5 workstations powered by the Intel Pentium 4 microprocessor. Though the initial investment while buying the 5 UltraSPARC’s is pretty high, but the picture becomes clear when we consider the whole scenario. Sun SOLARIS 9 running on these systems, with Sun ONE web server software configured is an ideal system. Support and maintenance can be assured by hiring a qualified Sun Certified SOLARIS Systems/Network Administrator. Thus, though the one time cost is high, but by purchasing commercial hardware running commercial software the long-term cost is stabilized and can be closely monitored. Moreover, regular patches, fixes and upgrades to all software from Sun can be monitored and applied as when required. On the other hand, Linux running on Intel Pentium 4 systems guarantee no support, no proper functioning of the software as well as hardware, no regular fixes, patches and upgrades to the system-level software, no maintenance of the application-level software and more significantly though the one-time cost while purchasing hardware and software may appear low, the long-term investment grows incredibly as time passes. Supporting Linux can be achieved by hiring qualified professionals well versed in all skills related to Linux. Unfortunately, certifications available are numerous and chaos still reigns supreme. A Red Hat Linux Certified Engineer (RHCE) well qualified in Red Hat Linux server administration would be pretty useless in a corporate environment which consists of 100 Pentium 4 workstations running TurboLinux. Linux distributions are numerous and so are the numbers of certifications available in the market. Even hiring professionals for the task can be a real challenge for the corporate manager. Support and maintenance can be a real pain-in-the-ass under such circumstances with a constant fear of losing acute company information, trade-secrets and invaluable client information always at the back of the mind. Big commercial systems dealing with terabytes of information or million dollar investments ought to put support, maintenance and stability of their currently-running systems first to anything else. What happens if something breaks? Another region of space where almost all free system-level software including Linux, FreeBSD, OpenBSD and others freak out is upgradeability. Nicely put, hardware systems running free software are not easily upgradeable, and even if they are, are not upgradeable to a great extent. Linux performs the worst in this case. FreeBSD on Intel platform though is horizontally upgradeable to some extent, but miserably fails (like Linux) when vertical upgradeability is considered. Commercial systems perform totally different, on the other hand. Sun SPARC and UltraSPARC systems are both horizontally as well as vertically upgradeable to a great extent. Thus, while upgrading such already existing commercial systems running commercial software, the amount of investment required is less and the productive output achieved is more. Thus, though Linux itself if freely available, many other aspects related to it are not free, and worse, pretty expensive and useless when compared to commercial systems. Linux may help corporate managers lowering down the one-time investment in hardware and software initially, but long-term investments get adversely affected thereby leading to a state of misuse of company resources, and more importantly support and maintenance of systems running such free software is never guaranteed, even if guaranteed by some firm, is usually of low quality and service. Corporate systems that prioritize safety and long-term investment costs to initially low one-time investments must stick to choosing commercial hardware systems running commercially certified software. About the Author Subhasish Ghosh, currently a final-year student of Informatics and Computer Science Engineering at the Moscow Power Engineering Institute (Technical University) tinkers with Linux, FreeBSD, OpenBSD, NetBSD and Sun Solaris operating systems. He has MCP, MCSD and MCSE certifications; currently preparing for Sun SOLARIS Systems Administrator Certification. For enquiries, comments and questions, send e-mail at: subhasish_g@rediffmail.com [Editor's Note: See Paul Ferris' editorial in response to this.] |
|
|
Comments (27)
Sun is better?
but it is clear to common woman
I'm rather flabbergasted by some of the author's assertions, such as "Linux when first appeared though did manage to make quite a good name for it, however failed to live-up to its reputation in the years to follow." When Linux was an infant, it was quite limited. It has grown consistently and impressively, and gotten better with every release. Also the author is not clear on whether he is discussing the kernel, or the whole Linux universe.
"Stable kernels appearing late, innumerable number of useless rewrites of kernel code in a few years, unavailability of proper marketing sources and failure to fulfill the predictions of the community has not only slowed down the number of users in the recent times, but has led to a more serious and deeper thinking about Linux's future in the recent years to follow." errr, wot??? The kernel is not on a release schedule, and never has been. When it's ready it's released. Slowed down the number of users? I guess all the large growth numbers I've been seeing are an illusion, then.
"it still lacks the efficiency of NetBSD when it comes to free portability." My jaw droppeth further. I'm unsure what this sentence means, though I do know for a fact that Linux runs on more hardware platforms than any operating system.
Skipping ahead to the conclusion in the last paragraph that Linux is more expensive long-term and unreliable is just plain nuts. Those of us out here in the real world, who have been administering Unix/Windows/Linux systems for years don't prefer Linux because of the cute lil penguin. We prefer it based on experience and real-world usage, which is something I doubt a college student has much of.
Folks are entitled to their opinions- I prefer some factual basis for opinions, thank you.
Actually NetBSD is the most portable operating system available. "It currently supports fifty four different system architectures, all from a single source tree, and is always being ported to more,"[1] however portability isn't just about how many architectures an operating system supports but, more importantly, how easy it is to port to new architectures.
"Due to NetBSD's ease of portability, Wasabi was able to get up and running on AMD's x86-64 platform several months sooner than other teams could port Linux."[2]
"According to Ed Gasiorowski, director of developer relations for AMD's Computation Products Group, the port to AMD's Opteron processor was extremely fast. 'I think the initial port was probably about two days [and] might have been as short as one day. That was a full 64-bit port'......the speed of the port was due to the quality of NetBSD's codebase."[3]
...And it's open source too:)
1. http://www.netbsd.org/Releases/formal-2.0/NetBSD-2.0.html 2. http://www.extremetech.com/article2/0,1558,1150126,00.asp 3. http://www.newsfactor.com/perl/story/21244.html
and look down about 6 or 7 links till you get here:
http://mail-index.netbsd.org/netbsd-advocacy/1999/08/27/0000...
And you will find me apologizing for writing an article on LinuxToday that claimed Linux was more portable :)
At least you didn't go *that* far :)
On a side note, though -- has anyone here had some stick time with NetBSD lately? Last time I tried it I found it to be a bit rough around the edges. It's not that I couldn't get some neat things done, it's just that I would have had a hard time giving a copy of it to anyone that wasn't a decent *nix jockey and a CS major, and that had a ready supply of coffee ...
Yeah, in other words, it *is* more portable, and that's a good thing, but it's not exactly in the same class as say, Fedora RC3 or Ubuntu, where a general user can easily run or maintain it... But maybe it's gotten better latey (I don't want to step out too far, I don't feel like apologizing again :)
--FeriCyde
I have generally found that the "less hyped" operating systems have been the one's that have provided me with the best and most reliable tools, such as NetBSD and Debian, but you are correct, I wouldn't recommend NetBSD to somebody who isn't familiar with the commandline and a text editor, or willing to learn (NetBSD's man pages are some of the best I have seen, by the way).
On that note I have found that OS's like the BSD's and Debian are excellent for learning unix skills.
NetBSD is pretty slick for sure, I've played with it a bit. I have no stake in Linux being the most portable, I am happy for NetBSD taking the prize. So I guess we give Mr. Ghosh credit for being right on one point. :)
However, Gentoo is not for the faint of heart. Getting the initial install done is a day or two process and don't even think of attempting it with a slow connection. Its default mode of operation is to download source code and compile everything (yes, everything) specifically for your hardware. On a slow processor, you could be compiling for days or even weeks (however, they have added a binary ISO based install.) This system is compiling the latest OpenOffice.org in the background as I type.
Its a long process, but when its complete you have a very fast system that's easily kept up to date with their portage tools (which, if I understand it correctly, is a knockoff of BSD's updating tools.)
Just FYI.
NoDough
A miserable effort
That was a downright miserable effort at misinformation you put forth there. While your premise - that "Free Software in Reality Isn’t Free $$$" (dollar signs mine) is nothing more or less than exactly what advocates of "free" software have been trying to get across for nearly 2 decades, your write-up is riddled with falsehoods, misperceptions, and misinterpretations. I definitely get the feeling that you are not exactly "in the loop" in regards to many of the topics you touch on, perhaps you are somewhat isolated in this regard due to your position at the Moscow University.
(1) "Linux when first appeared though did manage to make quite a good name for it, however failed to live-up to its reputation in the years to follow."
To put it kindly, Subhasish, what the **** are you talking about? I've got plenty of ammunition, but I'm not going to get into a market share argument with someone who displays such a massive ignorance of more fundamental realities. Open Source in general, and Linux in particular, has resulted in the most significant realignment of the computer business since Microsoft came on the scene. Proof is legion, but I will give one irrefutable example: the rise of Linux has produced what was once considered unthinkable, rapprochement between Sun and Microsoft. Your numbers, which are questionable in the first place, also don't tell nearly the whole story: Linux has snapped up market share as the various Unixes which populated the market throughout the 90's have gradually slid to irrelevance. On the other hand, and because of Linux, Sun has not made significant share gains in that market over the last 3-4 years. While you may not appreciate the significance of these facts, believe me, Wall Street does, as evidenced by Sun's horribly stagnant stock price. Do you even get the news over there? Have you read ANY of the analyses/breakdowns of Sun's latest quarterly reports? Let me give you one key passage from a piece at the Motley Fool (http://www.fool.com/news/mft/2005/mft05011402.htm?source=ept...): (snip) Operating income $16 million Interest income, net $33 million That's right. As hard as the company toiled away during the quarter -- the blood, the sweat, and the tears that fueled the server specialist -- it was shown up by being half as productive as the Sun's collection of idle cash and income-producing securities. (/snip) Or, how about this headline from CBS Marketwatch "Despite profit, Sun retreast as Q2 revenue disappoints" ([url=http://us.rd.yahoo.com/finance/external/cbsm/SIG=11g622ocl/*http://www.marketwatch.com/news/yhoo/story.asp?source=blq/yhoo&siteid=yhoo&dist=yhoo&guid={47E68D57-B0B4-4D1A-BD58-21B4B1227023}]http://us.rd.yahoo.com/finance/external/cbsm/SIG=11g622ocl/*...[/url]). Do these analyses mean anything to you? Or are your skills strictly limited to misinterpreting what misinformation you do have?
(2) Linux (also FreeBSD, OpenBSD, NetBSD) lags from commercial software like Microsoft Windows 2000, Microsoft Windows Server 2003, Sun Solaris, IBM-AIX and others when it comes to supporting the installed bases. Linux (and the community as a whole) provides no guarantee of support and maintenance to its users whatsoever. Once installed, who is in charge of proper configuration of these systems? Even once when properly configured, who is there to maintain and support these systems 24/7 around the year?
Pure FUD, Subhasish. Simple answer: whoever you have paid to do so. That issue has been thoroughly addressed, you are recycling a 1998 criticism.
(3) Nicely put, hardware systems running free software are not easily upgradeable, and even if they are, are not upgradeable to a great extent. Linux performs the worst in this case.
Again, WTF are you talking about? A completely unsupported assertion. My personal experience? How about this: when upgrading the hardware on an AIX server, believe me, you'd better either be a full-blown AIX guru, or have one sent from IBM directly on hand. I just don't know what you are talking about on the Linux side, since you don't even say.
(4) Linux distributions are numerous and so are the numbers of certifications available in the market.
Do you have a single clue about Linux? The vast majority of Linux distros are based on a fairly short list of other distros (RedHat, Debian, Slack, and in it's own category, SUSE). Even between those four, many skills are highly transferrable. All base knowledge is of course transferrable, and at this point, many key administrative utilities have been cross-ported.
I could go on, but why bother. The real question is, who are you and why did you feel qualified or motivated to pen this bundle of crap? You really need to get out and about a little more.
Peter Yellman
24/7 Support from Linux community:
Server Farms
I've experienced uptimes of over a year on machines configured as web servers running Red Hat and Debian. I have one customer whose website gets over a million hits per month.
I've had no trouble keeping software up to date on any of these machines and several times I've upgraded to new hardware. I relied on the server farm's staff to install the operating system and I moved over all my clients. Except for Rackspace, monthly costs are under $100 per server.
I do all the administration myself and when I started I knew nothing about how to administer a web server. I learned as I went along. I relied on news groups, forums, online documentation and the occasional O'Reily book.
- greg strockbine
Final year EGO!!
Debian, Red Hat, AIX, Sun Solaris Sys Admin. RHCE
--FeriCyde
Precisely. And sign your real name, like I do.
Precisely. And sign your real name, like I do.
Duh...
Current linux 2.6 is near of the number of platforms supported in netbsd if not more.
"Linux (also FreeBSD, OpenBSD, NetBSD) lags from commercial software like Microsoft Windows 2000, Microsoft Windows Server 2003, Sun Solaris, IBM-AIX and others when it comes to supporting the installed bases."
Wrong. IBM sells you support for Linux, Sun too. Sun has been selling linux servers for a while. So does HP. Those three sel you hardware + support.
"Linux distributions are numerous and so are the numbers of certifications available in the market"
So go and use redhat. Why would you use turbolinux, unless you're stupid? Linux is about choice. There are thousand of distros because there're diferent needs. For servers/workstations I recommend you redhat/novell.
"Big commercial systems dealing with terabytes of information or million dollar investments ought to put support, maintenance and stability of their currently-running systems first to anything else. What happens if something breaks?"
"Big commercial systems dealing with terabytes of information"? You mean systems like the 512-cpu 2TB RAM altix SGI sells with linux, and no option to choose other OS? If you want support you call IBM, SUN or HP and you use the support contract you probably have already contracted with them, because if you want support.....you contract it, no? i mean, were you expecting to being able to "download a support"? No, linux still don't gives you that.. :-)
"Another region of space where almost all free system-level software including Linux, FreeBSD, OpenBSD and others freak out is upgradeability"
I can run linux from a handheld device to a 512-cpu-and-docens-of-disks SGI Altix. That's a platform scalability you won't find in commercial unixes (they don't support handheld devices) neither in Microsoft product (biggest version of win 2k3 for itanium supports just two partitions of 64 CPUs each one, FAR from the 512 CPUs of a SGI altix running linux). And it does that in docens of hardware platforms, unlike windows (it just supports x86, opteron and itanium) or Sun (just x86 or sparc) or AIX.
Exactly where is linux lacking "upgradeability" in hardware? It must be in a windows for s390 (doesn't exists), solaris for itanium (doesn't exist) or aix for sparc (doesn't exist - linux will run ok on those three hardware plaftorms)
"Even once when properly configured, who is there to maintain and support these systems 24/7 around the year?"
Redhat/IBM/Sun/HP/Novell/SGI support and a engineer you've obviously contracted? I guess you've looked and considered all the different offerings there are TODAY (not 5 years ago, which is the enterprise world you seem to be describing) in the market from al vendors, for your enterprise, true " final-year student of Informatics"?
Really, this article is FUD in its whole. Sure, nobody gives you free support for what you download from kernel.org, _what_ made you think that, Subhasish Ghosh? Support is not "downloaded", it's "contracted". You don't get support using debian neither you won't find support using a cracked version of win2003 or solaris. Why on earth do you expect to get free support from OSS community? Are you aware that linux is the fastest growing OS platform on servers? That's probably because the people in charge of buying those servers actually have a _clue_ of how the world works, and buys support from bug vendors.
Unless you want to support yourself with your own engineers, which is what many people does and what you can't do with comercial vendors. Remember, Linux == CHOICE.
>And it does that in docens of hardware platforms, unlike windows (it just supports x86, opteron and itanium)...
Actually, Microsoft dropped support for the Itanium. http://slashdot.org/article.pl?sid=05/01/06/2248245&from=rss
NoDough
So many things wrong ...
In 1991 it was a fun hobby for a guy named Linus.
> Linux was initially designed to run on the Intel 32-bit platform.
No it wasn't.
> guarantee no support, no proper functioning of the software as well as hardware, no regular fixes, patches and upgrades ...
Well, if you download it for free and don't want to pay for support it is only supported by the community. As far as fixes, patches and upgrades ... thats all free and can be downloaded from the vendor. Red Hat has a yum update agent, Debian has apt, etc...
>No it wasn't.
I thought Linus started the project for 3 reasons:
1. He wanted a POSIX compatible OS for his 80386. 2. Minix was not capable enough. 3. He wanted to learn more about the i386 architecture.
-Steve
Also note that though it's a job and not a hobby Linus still has fun.
But man, what a serious hobby :)
--FeriCyde
The best always are.
Most missed analysis I have read recently...
I wan't to give some background on me - I am IT specialist with CS background - right now I work as IT Analyst and study more business oriented stuff (as I need to know business processes in my work). This may have no significance to you but I just to wan't to state my perspective...
You have made some points, name them:
1. Linux is not free (as in beer) since it requires to be installed, operated, supported by staff - yes, this is true. Nothing new here.
2. You can't get Linux with support contracts - this is not true - you can get support from vendor directly (think buying Red Hat, paying them additionally for support) or indirectly (think buing Red Hat, paying some other firm (probably local) to do the support). You can get that - just do some research - question is it cheaper? But you haven't answered this question you just lied that you can't get it in the first place. As for Red Hat put here any commercial Linux vendor (you are from Russia/Moscov so maybe ASPLinux would suit you - here you go: http://www.asplinux.ru/services/) or free as Debian - you can install Debian for free and then hire some external firm to maintain it. Question is cost.
3. You misunderstood numbers (as reffering to Solaris market share) - yes maybe it has not dropped down but it has not rised also. And Linux market share has rised significally (and Windows also) - replacing other unices. So if something is not growing (and there is a place for it) it is actually getting smaller as the concurence is getting bigger.
4. I don't like fact that you are stating some arguments as is - you don't explain anything but state f.e. "Linux kernel releases are late" or "Linux is unupgradeable" etc. - OK but *nobody* knows what you mean giving such opinions... Please reason them because now we may belive you or not - I don't belive you. Please explain to us what you meant by saying so?
5. You state that "Linux (compared to other commercial offerings) guaranties no support..." - OK Solaris has timeframes for support Windows has timeframes for support (in fact quite funny some people are now hot with Windows NT being unsupported soon) than Red Hat has support timeframes and also Novell does. So please tell us where is the difference? Also consider scenario (quite stupid scenario but read it) - Sun as a firm ceases to exists (f.e. they go bankrupt or bomb wipes its HQ out) - you are left with Solaris. And now the same with RH - you are left with RH - but with RH you got *source* *code* and that *is* *the* *guarantee* *of* *support*. I could go longer here but I am certainly sure that you will again loose a point reading it so I won't....
And the last thing - now I will make fun of what you have written, especially this quote:
> Purchasing commercial hardware that runs > commercial software is the best choice available.
LOL - what did you mean? I actually know that you meant hardware dedicated for platform (you can get nice multiprocessor Itanium servers designed for Linux) but where can I get uncommercial hardware (such thing even exists?) for free? It looks like your choice in fact is the only one... :)
Also saying that SPARC workstations will be *always* cheaper that Intel/P4 ones is like silly. :) Also considering Solaris as viable worksation OS is silly (if it is so good on workstations why does Sun sell Linux for Intels?).
And last word - considering support/maintainance cost please keep in mind that such cost are always in proportion to hardware and licenses costs. So when you have $5000 a month staff and $500 a month system - system cost is insignificant. When you have $500/m staff and $500/m system this proportion change. Now keep in mind that system (hardware/licenses) costs (be it any system) are static everywhere. Staff costs different depending on country (you should know that already - you Live in Russia, I live in Poland for that matter). So when you get cheap staff (like Russia, India) it plays less in this proportion than systems... I know you still don't get it.
Post a comment
A free account is needed to post. Write your comment here first: your text is kept in this browser while you sign up or log in, and is waiting for you when you come back.
Your message:Sign up and post (it is free) or Log in and post
It also doesn't lock you into a single vendor that can squeeze you for every penny once you are committed. I manage RHEL systems, SuSE systems, and a few Mandrake boxes that someone else set up with no problems. Linux skills transfer well between distros. It is mainly the package tools and a few default file locations that differ. The LSB is addressing these issues.
The only part that made sense to me was hardware scalability. But very, very few systems require "terabytes of information or million dollar investments". If you happen to get a job like that when you get out of school, good for you. In my 19 years of experience in this industry, I've run across very few of those.
IMHO, the value of Linux in performance, stability, and ease of management is far superior in many situations. Oh, I also have an MCP, MCSE, CCNA, and LPIC-2. Take a look at lpi.org for a good cross distro Linux certification.