Apples and bananas

Apples and bananas


While trawling through this week’s normal helter-skelter barrage of free software and open source news items, opinion pieces and analyzing ponderings a couple of pieces caught my eye. These are the BBC’s article entitled “Charity shuns open source code” and Silicon.com’s one called “CIO Jury: The Linux desktop is dead”. When first seeing these pessimistic pieces of free software doom and gloom, I confess my immediate reaction, as an advocate and developer, was one of misery, depression and fed-upness. Was it all worth it? What is the point? Where is the bright side? Should I simply go outside and step under a bus?

After a nice strong cup of coffee and pulling myself together a bit, I examined the articles a little more closely. I discovered that the authors, or originators, of each had, in fact, made a very common mistake while performing free and closed software comparisons that reminded me of the old adage regarding apples and bananas...

In each of the articles various scrutinies were made between the two contestants. In the blue corner was Microsoft and it’s Windows software, and in the red GNU/Linux and free software (maybe the other way around if you would prefer). The articles were eager to dive into a report of the fight between the two, but I would rather step into the changing rooms beforehand and examine the participants in a less distracting and more revealing scenario. If you will let me I would like to share my observations.

First, to examine the Microsoft camp. Microsoft is a closed software company that is remarkably successful. It sells a wide variety of software, but as far as I can tell its cash cows consist of sales of its operating system and Graphical User Interface, MS-Windows, and its office suite, MS-Office. It is very dominant in the desktop market, and Microsoft are using their top position in the market to try and maintain that dominance for as long as possible by employing whatever means it can get away with.

And now to the GNU/Linux room. Linux is the name a friend of Linus Torvalds came up with when asked to host a POSIX kernel that Linus had been working on. Free software predates this and is a method of developing and distributing software such so the end users and other developers have guaranteed freedoms of it and its derivatives, and “Open Source”, also mis-mentioned in the articles, is a type of certificate, or stamp, that software licenses can obtain that will guarantee the recipients, and only the recipients, of the software these freedoms.

Now, with that in mind, looking at those two champions in the ring, things look silly. There is no contest. This is not because one is stronger than the other, but because there simply is no match. Any survey that currently looks at desktop use will always find that Microsoft is way in the lead because they are the dominant players at the moment. If the surveyors looks at servers they may well find that software running on GNU/Linux is ahead. On a more obscure note if they searched for desktop adoption in the Administration offices in the City of Munich they would find that desktops running on the Linux kernel is the dominant choice, and that the Microsoft one is being dumped. Such surveys mean little.

In the CIO Jury article, their was an interesting quote attributed to Graham Benson of Play.com:

"Linux is a great example of the old adage ’you don’t get owt for nowt’. It is not free, as you pay for the support and there are so many flavours that it dilutes any potential attractiveness..

This statement is incorrect in almost every sense. To go through it point for point:

  • Linux is a great example of the old adage “you don’t get owt for nowt”: I assume Mr Benson is referring to the GNU/Linux distributions rather than the Linux kernel itself, and in particular to desktop distributions, nonetheless my reply concerns all these. Although GNU/Linux is free (see next point) no one is claiming that it can be up and running with no investment. Training is always required, as is getting to become familiar with the system itself, and all this costs money and resource. It is worth noting that new versions of Microsoft software require similar investments in time and money as well. This saying is a generality that applies everywhere irrelevant to the discussion.
  • [GNU/Linux] is not free: Actually it is. Both the GNU/Linux Distributions and the Linux kernel are free in almost all senses of the word. It is encumbered with no license fees or restrictions, and you can do almost anything with it when you get it. The only real restriction is that you cannot deny anyone else you pass the software onto the same freedoms.
  • [Y]ou pay for the support of [GNU/Linux]: I do not know about you, or Mr Benson, but I do not pay for GNU/Linux support as I support my sites myself. The fact I can pay for support if I want to hardly makes it less free, and the fact I can pay anyone I like for it, not just a single master vendor or an agent of it, makes it even more free.
  • There are so many flavours [of GNU/Linux] that it dilutes any potential attractiveness: What a curious thing to say. Taking that argument to its logical conclusion would be like saying a cafè should not sell ice cream because the attractiveness of that is diluted due to too many flavours available. Although I concede that different flavours of the POSIX desktops, namely GNOME and KDE, can give the illusion to software houses of complications (it doesn’t, but it can give that illusion) and put them off. The fact that GNU/Linux itself is so versatile and will run on anything from wrist watches to super-computers, I would have thought, would have made it more attractive to many more tasks. Also, to reiterate a previous point, new versions of the Microsoft operating system can cause just as great headaches.

Mr Benson assumes that the lack of penetration of the GNU/Linux desktop he observes is due to technical inferiority of the free systems. The fact of the matter is that the free offerings are not significantly less advanced, and are in many aspects more so. The figures have more to do with people’s unwillingness to part from what they see is the norm. Despite that, GNU/Linux adoption, including desktop inclusion, is on the increase. When it has reached a certain percentage—and it is when, not if—then you will see the system deployment analysis change accordingly.

The criticisms of GNU/Linux in the BBC article bear an even more distant relationship with reality. The crux of the argument of Mr Steven Buckley of Christian Aid, is that it was pointless for him to use free software solutions when Microsoft gives him licenses for their software for practically nothing and that it is easy for him to find expertise in it. I can accept that, and looking at the short term for his charity I can accept the logic, and I would like to take this opportunity of thanking Mr Buckley and all the people who work for Christian Aid for the marvelous work they do.

The problems I have with Mr Buckley is the arguments he gives against free software that show a remarkable ignorance of it, and he may well be missing an opportunity to save money. First of all, he has the same misconceptions of free software support as those made by Mr Benson above. He incorrectly assumes you have to pay for support if you use GNU/Linux. This is not the case. And, in fact, I know of charities that use GNU/Linux and who receive free support from IT professionals as their contribution to the charity. I do not believe Christian Aid will find support for GNU/Linux any greater—and perhaps less—in cost than that they currently receive for Microsoft solutions.

The second misconception he has is in understanding the mechanics of free software development. To quote the article...

He also explained that what is seen as one of the advantages of open-source - that the core code can be examined by anyone - could actually work against the charity.. "We are a funding organisation that ships £90m [UKP] around the world - the last thing you want to do is open up your systems to anybody to have a look at to deal with bugs," he said..

Please let me point out the error of the above supposition. Nobody is suggesting that he should open up his systems to the world if he uses free software. In fact, no one is suggesting that he should open up his free software either. Bug fixes, should they be needed, are usually distributed by the system maintainers for the package that they occur in, no matter who they are fixed by. And, they can be deployed on any system without the need of opening it up. As for the 90M UKP is concerned, Mr Buckley needs to understand that free software developers take security very seriously. The openness of the software means that developer peer-reviewing of the software’s reliability and security results in an exceedingly rapid discovery, diagnosis and fixing of such problems. In my experience, this results in far more secure and reliable system functionality than is offered by Microsoft and other closed software companies.

Using discounted Microsoft software may well make short term sense for a charity, but I doubt it makes long term sense should all the requirements be properly analyzed.

I am however digressing. The discussion should not be about a fancy tag fight between GNU/Linux with free software versus Microsoft with closed software. As mentioned at the beginning of this entry the two sides of that are very different things anyway. This is about correcting misconceptions in the hunt for the best IT solution for a particular IT problem and free software’s role in that.

When discussing the modern POSIX desktop solutions, the assumption normally is to think of GNU/Linux. However, the desktop using GNU/BSD, GNU/Mimix, GNU/Solaris, GNU/HURD or any other GNU/Name-your-own-POSIX-system (even closed ones like SCO) can be used (often with such similarity that you cannot immediately tell which system it is). It’s even possible to run one under MS-Windows using the GNU/Cygwin layer, even though at a lesser extent. This is possible through the POSIX and the free software community’s insistence on developing and following strict standards to allow not only interoperability but also component replace-ability. Should you not be happy with the operating system you are using you can replace it without breaking all your applications. With free software, lack of choice and competition are a thing of the past.

Although Mr Buckley mentions that Microsoft gives him very good deals at the moment, can he be assured that they will in future? Bill Gates is retiring, what if Microsoft starts losing money (can always happen) and Bill’s successors are not so charity conscious? The corner, known as “Vendor Lock In”, which Mr Buckley is painting his charity into at the moment suddenly does not seem to be that attractive.

Even if the Microsoft solution is chosen, that does not disqualify the free software solution. OpenOffice.org Office Suite, GIMP Image manipulation program and a host of other programs that largely make up the free GNU/Linux desktop solutions are also freely available for MS-Windows. These use the open standards available to all as well. This means that, even if MS-Windows is insisted on, you can still benefit from free software for your IT solution. It does not take a genius to see that blindly discounting free software as unimportant, while designing IT strategy, would almost certainly cost you money.

There is another problem with blindly following a single closed vendor for all your software, you lose the ability to perform genuine comparisons of offerings. GNU/Linux, by its nature, naturally supports thin-client technology out of the box. At a push, it can support Multiterminal technology, where several monitors and keyboards can be plugged into a single PC producing a genuine multi-user box. Use of this technology would enable re-use of old and otherwise obsolete hardware (as the thin clients) and also enabling consolidation of hardware resources producing capital investment savings. For a charity whose IT requirements require distribution of 90M UKP, this would result in quite a saving due to not having to purchase expensive hardware. Also, should one old thin client fail, all you need to do is replace it with another. Making and restoring backups for the physical desktop machine is not an issue as no data or programs are stored there. This would save significantly on hardware support. The software on the centralized application servers would also be a lot easier to maintain as data and applications are not sprawled over countless desktops but centralized on a finite number of controlled servers. This means that more money is saved regarding software support too.

The thin clients scenario is only one example of many I could rabbit on about. There is also the ability of using lightweight desktop software like XFCE that will also increase the use of legacy hardware further. Also on the same note, lightweight free office programs such as AbiWord and Gnumeric that use the same file format as the OpenOffice.org suite so they are truly inter-operable with each other. There are many other features I could point out as well, but unfortunately it is late and well past my bed time, so I had better start drawing to a close.

I would have thought the advantages of those things alone would warrant the attentions of system administrators of a large charity, and I have not even begun to mention the potential of tailoring the software itself to exactly fit the requirements wanted.

In conclusion, comparing Microsoft/Closed-Software/Locked-In Support with GNU/Linux/free software is like the old adage of comparing apples and bananas. More to the point, if you follow the closed Microsoft route exclusively you find yourself with a metaphorical, say, orange. It may be a big juicy orange, but it still is only an orange and will never be more than that. With the free software route you not only end up with an orange but also bananas, pineapples, lemons, limes, pomegranates, durians (yummy!) and anything else you may want. The choice may be initially confusing, but how many restaurants do you go into that only offer one choice of meal? Why limit yourself this way when it comes to software?

Goodnight.

Category: 

Comments

Anonymous visitor's picture
Submitted by Anonymous visitor (not verified) on

I own several franchised fast food restaurants. One of the things I have learned, and our franchisor preaches to us, is to avoid the problem of having too much variety. Too much causes confusion for the consumer. Imagine walking into your favorite place and looking at the menu. Some are a bewildering area of choices. It makes you just skip past all that and order the same thing you always did. If you had never been there before, you may not order anything and leave because you can't figure out the differences or don't want to take the time. There is no familiarity or comfort area.

The same can happen with Linux. I am only now trying this OS but spent too much time trying to figure out which distribution I want to start with. When I tried looking into the *BSD OS, the choices were much easier since there are only three.

So you can narrow OS choices down to Windows, Linux, *BSD, Unix. Unix is expensive, *BSD is not known for the desktop, Linux is splintered and Windows is a singular concept.

When it boils down to the bare minimum, though. Linux is still Linux and a singular concept. What Linux should try, perhaps, is advertise Linux first, then clearly show the different flavors. Sell the ice cream first, then show the flavors.

Anonymous visitor's picture
Submitted by Anonymous visitor (not verified) on

I'm glad that when I go to buy a car I have multiple choices, and have the same now when looking for a new computer system. The fast software Big Gates shops all have the same packages, although even these have several dozen flavours, but when I visit my frienly Linux retailer, I can get one custom built for less than a Big Gates.

Like the article says, unless you are addicted to the palm sugar in the Gates package, you can get a diet to fit your real needs.

Anonymous visitor's picture
Submitted by Anonymous visitor (not verified) on

At my current workplace, we have 20 users who run Windows on thin clients. The thin clients don't actually run Windows. They run Linux. They boot Linux off a server first. When the Linux boots, it runs 'rdesktop', a free remote desktop client. Rdesktop then connects to a Windows 2000 Server running Terminal Server. The Windows 2000 server is actually a virtual machine, running under the free VMWare server, on top of Slackware Linux.

To the end user, all this is invisible. They see only Windows. To the system administrator, his job is easier because he has only the Windows 2000 server to worry about. Any changes/updates are instantly available to all 20 users. Less downtime, less hassle for everyone.

To the finance director, my boss, the beauty is the lower overall cost. Some of the thin clients are, in fact, old PCs. So we saved on the cost of buying new PCs. The others are from Diskless Workstations (http://www.disklessworkstations.com). No hardware issues, no compatibility issues.

Because the Windows 2000 server is running on a dual Xeon server with 12GB RAM, the end user feels greater performance than s/he would using the old recycled PC.
Everybody wins.

I can't imagine a charity not being interested in a win/win scenario like this. If I were the chairman of the board of that charity, the IT manager would be out of a job.

Seth Brown

Anonymous visitor's picture
Submitted by Anonymous visitor (not verified) on

There are days, especially where the world is going, where the life of the sanyasi seems positively attractive, others when you feel the need to get stuck in and do something good.

Anonymous visitor's picture
Submitted by Anonymous visitor (not verified) on

Is the polarised argument Linux vs Windows? Or is it Unix vs Windows?
If it is Unix vs Windows then the clear winner is Mac OS X.
It has that comfortable homogeneous UI that's easily supportable and all those lovely Unix underpinnings for geeks to play with.
It comes with GCC 4.0 and X11 and all API documentation on the install DVD (yup you can code to the metal out of the box).
It comes free with that sweet looking (subjective) Apple hardware (which will also run Windows and Linux if you really want to)
Updates are managed automatically.
Installing new software is simplicity itself.
You don't have to be a geek to use it.
All your existing source will likely work with ./configure; make; sudo make install (if you have installed any extra libraries required with fink/darwinports).

I love unix as a server OS (running several hundred RedHat boxes) and I love unix as a desktop OS (running an iMac 20" and a MacBook Pro). I used to work hard to make laptops work with Linux and use all their features, but I grew tired of having to do it with every new laptop I bought. I'd just about get it configured by the time it was obsolete and I wanted a new one.

I could support my mum running the mac of her choice and assorted software over the phone. I'm not certain that she'd follow my attempts to support her running Linux even if I could intuit which window manager, desktop manager, choice of buttons/taskbar she'd decided to run (and know the default hotkey combinations)... and she's not ready for the CLI. I had enough problems with Windows when I found she'd dragged the taskbar to dock on the left hand side of the screen over the phone.

Not everyone is a computer geek. Or wants to be. Those of us who are.. really appreciate OS X: its real beauty lies under the hood (although the bodywork is pretty slick too!)

Faye Pearson

Author information

Edward Macnaghten's picture

Biography

Edward Macnaghten has been a professional programmer, analyst and consultant for in excess of 20 years. His experiences include manufacturing commercially based software for a number of industries in a variety of different technical environments in Europe, Asia and the USA. He is currently running an IT consultancy specialising in free software solutions based in Cambridge UK. He also maintains his own web site.