daniel hertz

Why does the world run on Linux, even though no one uses it?

Linux is free and open source software(FOSS). All FOSS developers share a philosophy that software should be free - as in freedom, not as in beer. This means it should be available for all to use as they wish, to edit for their own purposes, to share (either in its original or its modified form) with anyone, and to learn from (What is Free Software? 2010). This philosophy shapes Linux, as it does all FOSS. In this paper I will ask whether this philosophy has also shaped its use in particular I will focus on whether there are differences in the uptake of Linux in the Home market and Business market.

I will use 'Linux' to mean the collection of UNIX-like operating systems that all use the same kernel. Its origins are typical of FOSS. Linus Torvalds (What is Linux? 2007) created the kernel in 1991 to replace MINIX, an education-only OS that lacked many useful features. His goal was a terminal emulator to control his university's UNIX cluster and support the full feature set of Torvalds' new computer microprocessor (unlike MINIX, which only supported the legacy features). Torvalds realised that he could create a fully fledged kernel from his terminal emulator, and developed it into a piece of software that manages all low level functions - such as memory and files, addressing cpu tasks, and communicating with hardware (Torvalds, 2002).

When close to completion, Torvalds posted his intent on the internet, and a number of other developers helped port many components from the free operating system project, GNU. Started in 1984 by Richard Stallman, the GNU project began the free software movement. It wasn't a new idea. Before software was granted copyright protection in 1980, users, mainly college professors, shared and edited software. GNU was started to show that users should still be allowed to share and change the software they own. Stallman's project produced many vital parts of an OS, but had yet to finish its kernel, (What is GNU? 2010) leaving the need to find a fully functional, and 'free' kernel. The collaboration led to the rapid release of Linux 0.0.2 and the subsequent expansion of the developer base. In March 1994, version 1.0.0 was completed, marking the stable kernel's release. (The History of Linux, 2008). The free software roots and philosophy of Linux were evident from the beginning, with Torvalds releasing the source with the kernel for free. Subsequently, as the community has grown, more and more free pieces of software have been produced for Linux.

Unix-like OSes all use the same modular structure; so even if all Linux and GNU-based systems have the same kernel, the X window system, and many of the same GNU parts, they can still be very different. Using desktop management solutions such as GNOME, KDE or another GUI, package managers such as yum or APT, and many other groups, edit the kernel to enhance it for their specific purposes. All these different versions are called distributions. Each distribution can be tailored for a specific purpose, such as running the most powerful computer in the world (the IBM Sequoia), or rejuvenating an old and sluggish computer (Damn Small Linux), or running a TV or a TomTom GPS. All this diversity is important to the Linux community. If each group did not develop, release and maintain its changes, that area of specialised development would be neglected and the whole Linux project would be more limited. All the Linux licence requires is that developers publish their modified code so everyone may learn from it, and that they do not charge for anything containing free code.

This has been an enormously successful model for software development in some areas. There are estimates that 60% of servers (Ballmer Still Searching for an Answer to Google, 2008), 89.2% of supercomputers (Linux: It doesn't get any faster, 2009), and 6 of the ten top hosting companies use it. But Linux only holds 1.09% (Operating System Market Share, 2010) of the home computer market.

Later on, this essay will focus on the use of Linux in 'the cloud', where internet services such as Google are hosted. In the cloud, users have no responsibility for the maintenance of the hardware or software, they just need to be able to connect to the services available to them. Most of the industry acknowledges that computing power is moving from the user to a server room far away (What is Cloud Computing and how do I use it? 2010). Linux has a huge market share in this rapidly expanding, vitally important but inconspicuous area of computer use, but users know very little about it.

At home, Linux finds its place on only about 1 out of every 100 computers, OS X on 5, and Windows on 94 (Operating System Market Share, 2010). At first sight, this is puzzling. Linux is free, while both of the major alternatives are very expensive. Users of Linux save £80-£140. As home computers have become ever cheaper, the OS can now be the most expensive element. Yet Linux makes little impression on the home market. I see four important reasons for this phenomenon.

The first is marketing. Apple and Microsoft have spent $2 billion to sell their wares to the home user (Apple's 2009 ad budget: Half a billion, 2009). Linux can't afford a campaign to change its image as nerdy, hard to use, and incompatible with everything. Moreover, consumers do not have a choice when buying the average computer - manufacturers, who usually profit from selling the software, preinstall the OS, giving users no alternative.

The second problem is resistance to change. Even if most users knew about Linux, they would not choose it. They would not want to spend time and risk breakdowns to learn a new system. , This inertia is powerful, even though many of the UI traits and metaphors carry over all three operating systems. Distributions like Ubuntu have a UI perfect for beginners; in fact, it is probably better laid out and more consistent than any Windows version. Remarkably, users are loyal to Windows and Mac OS even though they have both changed considerably and required users to relearn their software when upgrading. The problem here is that Linux was late to the modern OS party, so it was unable to create the user following that Windows has. Even though Apple was first to market, Windows 95 was what the user wanted, at the right time. As computers got cheaper in the late 1990's, it provided easy internet access, and worked on any IBM compatible PC. At that point, Linux hardly had a GUI.

Another problem is customer support. Users feel that an expensive operating system is more likely to be free of bugs and to provide assistance for users in difficulty than a free one. They value the support structure that they believe only bought software will give them. The ideology and philosophy of the software does not concern or even occur to them. In fact, most Linux distributions have an exhaustive help database. Thanks to volunteer support on internet forums, it's easy to get help with Linux from a human. But since Linux does not come with any warranty, there is no one to blame if it fails, and no financial compensation for any loss . Users do not believe that such an informal system of support can be effective.

The last problem is that no large computer producer cares about the Linux desktop. Hackers alone have created it - as an OS for themselves. This makes Linux seem rather hostile to a novice (though this is changing). If a company with a large consumer presence were to uphold its place in the Linux kernel, maybe it would have more chance. In contrast, customers in the server sector are used to versions of the kernel customized by the hardware manufacturers to suit their machines and their customers' needs, and then those modifications being submitted upstream. Currently, no major manufacturer is doing this for Linux as a primary OS. Companies like ASUS are creating ÒQuick StartÓ OSes as a secondary system, but customers cannot see the advantage when they can load Windows, instead of having to reboot when they want to use anything more advanced. To gain a presence in the market, Linux needs a large manufacturer to hire developers to optimize the kernel for the desktop space, as IBM has done with servers. The modifications would be sent upstream . This would give Linux the credibility it needs with domestic users.

New developments in personal computing may increase Linux's share in home use. The Smartphone & mobile embedded device markets are less dependent on old UI metaphors than the personal computer. Some manufacturers have developed their own proprietary OSes, but many manufacturers have produced free alternatives using the Linux kernel. Nokia's Meamo and MeeGo, Palm's Web OS, Google's Android, Amazon's Kindle and Motarola's mobile phones all use it. Apple's iOS, Microsoft's Windows Mobile 7, Symbian and Blackberry OS do not. This market may be where Linux can prove itself to the domestic consumer. Even though Linux-based systems are relatively new to the market, they have gained massive sales. Google's Android leads the way, gaining 8% of the market in a single year, while Sybian and BlackBerry OS fell (Gartner Says Worldwide Mobile Phone Sales Grew 17 Per Cent in First Quarter 2010, 2010).

This trend shows that, with significant advertising and large companies backing it, Linux can be a viable competitor in the desktop market. It may see this success later in the year when Google launches Chrome OS, its desktop operating system based on Linux, designed specifically to take advantage of cloud computing. As mobile embedded devices continue to take over the functions of the desktop, Linux could be in the space worth fighting for.

The situation with business users is very different. Online businesses use Linux for a number of purposes, whether they care about the philosophy of their software or not. It could be simply easier to deploy and manage, or the price of free is a great incentive.

Startups and small technology companies often have the knowhow to do many of the jobs of IT staff, but since they are also running a business, they don't have time also to learn lots of new software, or the capital to invest in expensive hardware and software. Linux meets their needs by being easy to use, cheap and secure. It is based on UNIX, is free, and runs on almost any hardware. It also works straight out of the box as a webserver, one of the most common tasks needed for a small business, or IT infrastructure. Linux appeared at the perfect time for all these factors to meet, since every company was scrambling for a website when the internet came into homes in the late 1990's. Startups at that time - such as Google - used it, and as the internet has grown, many of these companies have grown with it. They have become reliant on it, so even though they chose Linux for reasons which no longer apply, they stick with it. Moreover, at the time cheap hosting was not available, but the internet was spreading into offices, schools and workplaces, so larger companies also used Linux to host their own websites because it was cheap and quick to implement.

These companies not only use the kernel, they continue to develop it, as they depend on it being fast, stable, and responsive to new conditions. The growth of the internet and decrease in price of computing power has had a similar beneficial effect. Hacking away at the kernel has become a hobby for many. By 2000, powerful computers with internet connections were inexpensive enough for amateur Linux enthusiasts to buy a separate hobby system, speeding up the development of the kernel even more. This in turn has made Linux even more attractive to corporate users.

From the start Linux has had many advantages over Windows NT as a server OS. When the NT kernel was released in 1993, it supported the same architectures as Linux, but its main interface was a GUI, which requires more processing power and a way to see the output. On the other hand, a UNIX system like Linux allows the user to log into a terminal window remotely (Rau, 2010). By the time that everyone needed servers, Microsoft had developed NT further to run on Intel hardware that was cheaper than the bundled Solaris/UNIX software/hardware bundles at the time (Hall, 2010). But using NT meant that IT staff had no control over the software. It was proprietary, not FOSS. They couldn't quickly patch it if a security threat came up - Microsoft would need to, and staff would need to learn a whole new OS. Linux entered the consciousness of IT professionals at exactly the right time. Linux was new, free, secure enough, but they didn't need to relearn everything they already knew. But Microsoft had marketing money and corporate swing with a great many companies and developers, so many decided to use the Windows systems, and developers decided to create applications for them. This is why NT is now a market leader in server operating systems.

Also at this time IT infrastructure could be extremely varied within a company. Many large companies were contractually obliged to make their computer purchases from several different sources, which produced incompatible hardware (Hall, 2010). Since Linux ran on all these systems, applications s could be compiled in one location for all architectures. These applications were not optimally efficient, but that disadvantage was outweighed by the significantly shorter development time.

Eventually, large server companies, who before were only interested in selling their UNIX flavour with their hardware, started providing Linux-based software as well. IBM started supporting Linux in 1998, and has over 300 Linux Kernel developers to support their Linux server sales. Oracle (the world's largest database supplier) began to create software for Linux in 1998, and released their own Linux distribution in 2006. These huge companies rushing to support Linux demonstrated that it was a viable platform for business - and the support problem was solved. Now businesses could buy Linux servers with confidence: if they broke someone could be held responsible. This gave Linux on the server legitimacy and marketing dollars.

A great advantage of Linux is the ability to edit it to do what a system administrator needs. Skilled technicians can edit the code to do what they want, or ask someone else to do it (probably for free) if they can't. Contacting a giant software house to ask for a feature is nearly unheard of, as there is little or no chance of the feature getting incorporated promptly. As the source code is not editable, nasty workarounds can be needed to adapt the system to specific users' needs. The flexibility of Linux can be especially useful in creating new cloud applications, as the whole system can be optimised for a single task.

Also in the cloud, Linux has a great deal of scalability, as it has been built for huge systems like the IBM Sequoia. This makes running cloud services an awful lot easier, as large clusters can run applications used by huge numbers of people. These systems can even scale themselves if they are virtualized, scaling back the power of the system at off peak times. This is how Amazon Web services work (Amazon EC2, 2010) - Amazon sell their spare 'clock cycles' and hard drive space to anyone who wishes to use them as cloud computing, when they are not busy. This ease of large-scale hardware management is a key to cloud computing.

Linux is also very adaptable - it's easy to create virtual servers on top of it or as the OS inside a virtual machine (Gillen, 2010). This allows one server to host lots of websites or services in complete autonomy, so hopefully they don't all become unavailable due to a software fault in one of them. The autonomy requires at least two OSes - the guest OS, and the host OS. Linux can act as both, having a small enough memory footprint and enough features to be used as the virtualised OS. Two of these virtual servers can be kept in perfect sync, so a cloud service can reduce lag if there are lots of servers near its users. If Windows is needed, it can be virtualised as well on top of Linux for greater hardware flexibility.

For all these reasons, Linux is used for slightly different purposes than Windows. A larger percentage of Linux servers host web services and are used for research and development, and less for IT infrastructure and business processing (Gillen, 2010). Linux is more successful in the server market because administrators have more technical ability to take advantage of its open source nature. It is based on something they know, UNIX, has a very low acquisition cost, has become very well supported by hardware companies, and is very good in the cloud. Also it became available at a crucial point in the development of the computing industry - when UNIX was expensive, and Windows was new. I believe that this was the driving factor for Linux in the server market.

On the other hand, at home Linux is apparently very rare, mainly for a similar reason: it wasn't available at the right time - when computers were becoming cheaper. Linux on the desktop also has a less formal support structure, fewer developers and less marketing power than Windows. These two problems could be solved by a large PC hardware company championing Linux - providing support, marketing and developers.

But this rarity is only apparent: in reality, we are all Linux users. As such a large percentage of the internet and business IT infrastructure relies on Linux, the average consumer uses Linux every day - when we make use of Google's many services, such as Search, GMail or Maps, when we shop on Amazon, when we login to our employer's cloud at home. We all use it without even noticing.

All in all, humans are not rational beings, and we don't always choose everything for the reasons we should. Computer users are as irrational as anyone else, and they don't always choose software wisely. They stick to brand names they recognise or accept what appears to be urban myth. Users need to be told that there is a cheaper, easier way to do things; they need to be persuaded that free things can be valuable and reliable. To achieve this, a huge marketing drive would be required - similar or on larger scale to Apple's 'Get a Mac' campaign. But there is no company willing to do this right now. There is hope for Linux in the consumer domain though, as the industry starts to move away from the traditional desktop model - Google's Chrome OS could change the use of Linux, and all of computing. There is also significant presence in the mobile and embedded markets. This could grow from here even if consumers do not know that they are using Linux. Eventually the diversity offered by Linux could mean that we spend most of our time using it - in cloud applications, on mobile phones and other embedded devices, or even as an operating system. To sum up, Linux is handicapped with domestic users to some extent by its FOSS character and philosophy, though other factors may be more important in explaining its small share of the domestic software market. Among professionals, however, the philosophy is no disadvantage - in fact, it provides a flexibility which makes Linux more attractive.

(02/03/1999). IBM launches biggest Linux lineup ever. Available: http://web.archive.org/web/19991110114228/http://www.ibm.com/news/1999/03/02.phtml. Last accessed 01/09/2010.

Torvalds, L, Diamond, D (2002). Just for Fun: The Story of an Accidental Revolutionary. USA: Harper Paperbacks. p60-63.

Charlson, J. (24/05/2006). IBM Invests US$2.2 Million to Expand Linux Technology Center in Brazil. Available: http://lwn.net/Articles/185602/. Last accessed 01/09/2010.

(02/07/2007). What is Linux? Available: http://www.linux.org/info/index.html. Last accessed 06/07/2010.

Niccolai, James. (26/09/2008). Ballmer Still Searching for an Answer to Google. Available: http://www.pcworld.com/businesscenter/article/151568/ballmer_still_searching_for_an_answer_to_google.html. Last accessed 05/07/2010.

(2008) The History of Linux. Available: http://www.linux.org/info/linux_timeline.html. Last accessed 04/07/2010.

Vaughan-Nichols, Steven J. (23/06/2009). Linux: It doesn't get any faster. Available: http://blogs.computerworld.com/linux_it_doesnt_get_any_faster. Last accessed 05/07/2010.

Elmer-DeWitt, Philip. (28/10/2009). Apple's 2009 ad budget: Half a billion. Available: http://tech.fortune.cnn.com/2009/10/28/apples-2009-ad-budget-half-a-billion/. Last accessed 05/07/2010.

(11/02/2010). What is cloud computing and how do I use it? Available: http://www.independent.co.uk/life-style/gadgets-and-tech/news/what-is-cloud-computing-and-how-do-i-use-it-1896368.html. Last accessed 04/07/2010.

(19/05/2010). Gartner Says Worldwide Mobile Phone Sales Grew 17 Per Cent in First Quarter 2010. Available: http://www.gartner.com/it/page.jsp?id=1372013. Last accessed 05/07/2010.

(23/06/2010). What is GNU?, What is Free Software? Available: http://www.gnu.org/. Last accessed 4/06/2010.

Rau, A. ---[at]gmail.com. What does Akamai use? (29/06/2010)

(05/07/2010). Operating System Market Share. Available: http://marketshare.hitslink.com/operating-system-market-share.aspx?qprid=8&qptimeframe=M. Last accessed 05/07/2010.

Gillen, Al, Linux Server adoption: Where are we now? (public), LinuxCon 2010. (11/08/2010)

Hall, J, Open Cloud Summit (public), LinuxCon 2010. (08/08/2010)

(2010). Ubuntu | Dell. Available: http://www.dell.com/content/topics/segtopic.aspx/ubuntu?c=us&l=en&cs=19. Last accessed 05/07/2010.

(2010). Amazon Elastic Compute Cloud (Amazon EC2). Available: http://aws.amazon.com/ec2/. Last accessed 01/09/2010.

Why does the world run on Linux, even though no one uses it? by Daniel Hertz is licensed under a Creative Commons Attribution-NonCommercial-ShareAlike 3.0 Unported License, 2010

daniel hertz

about

my photos

Why does the world run on Linux, even though no one uses it?

my tweets