Its always interesting to do a Dave Letterman about Linux. You know, a top ten list of at least interesting if not funny things to say about the operating system a Swede from Finland built. But what about Dave Letterman 2 times over? Specifically this overview will proceed with Top Ten Facts about Linux and The Top Ten Problems Facing Linux and then try to carve out a story and overview around those Top Tens. Hopefully this will not be a (s)mashup like some current SaaS offerings or so confusing or tenuous in its many connecting dot.points that even Condi, Donald and Dubya would again be hopelessly unable to "connect the dots".

The Top Ten Facts about Linux

1. Google has been able to deliver massive, clustered, on-demand, Linux-based computing as a necessary (but not sufficient condition)to gain decisive competitive advantage in its markets;
2. 64% of Servers on the Web use Linux even though Microsoft has been able in March 2006 to switch 6% of all parked domains to Windows by having registrars become Windows websites. Before the switch Linux was trending above 70% market share according to NetCraft.
3. Embedded OS has seen Linux go from 37.5% usage to 42.7% between 2004 and 2005; while Windows and all variants have have gone from 13.3 to 15.6% in the same period. DOS and VxWorks are the big losers in the same period declining by about 2% each. Results are from Linux Embedded Devices survey of over 750 executives.
4. Gartner in 2004 expected Linux to grow to 1/2 of all servers by the end of 2008 - I wonder what they think now ?
5. Linux has over 300 distributions of which the top 10 have over 90% market share.
6. Linux runs on a PC with less than 1/2 the computing power required by Windows XP to run with good response time. With Windows Vista this gap will widen even more as Vista will have requirements for 1GB of memory with a 2.4GHz+ processor and a 32MBvideo card.
7. Linux runs on the desktop with all the GUI features of a Mac or Windows system as seen in programs like Gimp on Linux=Photoshop on Mac and Open Office or Koffice on Linux=Microsoft Office. The response time, ease of use, and ease of learning of these programs are comparable to their Mac or Windows equivalent. These GUI features include all the menus, widgets and GUI components one uses on a Mac or a Windows PC including drag and drop, right mouse click, and sophisticated mouse gesture recognition and operations.
8. Linux on the desktop for developers (with Suse Linux for example) delivers free on the distribution DVD compilers(C/C++, Fortran, Java, Perl, PHP, Python), databases (MySQL, PostgreSQL, Sleepycat, etc), application servers (Apache Web server, Jakarta-Tomcat and JBoss application servers, etc), top notch IDE development environs ( Eclipse, Mono-devel, Netbeans, etc) and dozens of other free open source or freeware developers tools. These tools can be immediately installed by choice and can save hundreds if not thousands of dollars over their Microsoft equivalents.
9. Linux on the desktop and as a server has had a better reliability and security record than Microsoft Windows desktop and servers systems for the past ten years. Currently Linux on the desktop+its browser (choose any of Firefox, Konqueror, Opera) is more reliable, secure and Web standards compliant than Windows + Internet Explorer and has been for ten years also. In sum, on the desktop and as a server, Linux has a distinctly better reliability, security and Web standards compliance record than Microsoft and its Windows line of products.
10. Linux for desktop and server is open source. One gets the source code with the distribution. One can change the source code and recompile the system. There are books available describing the process and how-tos. Microsoft Windows is a proprietary OS. One, under special circumstances and restrictions, can get some read-only access to the code. One cannot modify or change the code.

The Top Ten Problems Facing Linux

1.How does Linux dispel the image that commercial software and vendors are second class citizens in the Linux world ? Or alternatively how does one change the expectation that on Linux all software should be as close to free as possible ? Do some Linux distributions want to preserve "free for all" ? Will Linux ever be able to take the crown - the most programs and applications exist on Linux ?
2.How does Linux preserve its independence and innovative diversity with literally dozens of variants and distributions while at the same time providing enough core basic standards to meet the demands of massive markets. Can so many distributions intelligently vie for diverse segments like thin client PC desktop usage or for basic device markets of embedded computing, or for cluster farm server utilities ? Or do they step all over each other and not really match any one segment well ?
3.How does Linux face the OS-Lord of the Rings conundrum - "one Ring to bind them all, one Ring to rule them all". Given that the scale of computing devices needing OS support ranges from pico-like embedded devices with minimal memory, diverse instruction sets and/or massively varying environmental operating conditions up to the biggest Beowulfian clusters with huge multi-core devices and an army of diverse support servers - how can the core Linux kernel serve these divergent needs in a common way without splintering or becoming too huge and unwieldy?
4.How does Linux match the marketing muscle of Microsoft with "paid for by Redmond" technical seminars, research studies and a website widely advertised on the Web and in print that purportedly Gets the Facts on Windows and Linux comparisons. But this same site pointedly omits key issues that would change the balance. No magazine or print website has dared to challenge these "Get the Facts" studies, cases and white papers - and why should they bite the hand that feeds with some of their biggest ad revenues ?
5.How does Linux meet demands of specific markets such as multimedia or storage management that look for specific hardware+OS modules that serve their needs but may even hinder or degrade performance and/or reliability in other segments ?
6.How does Linux cope with governments that want to impose their security, availability and privacy access rules, filters, and routines onto Linux apps and even into the Linux core and kernel? What would happen for example if China or the US NSA insisted on creating versions of Linux with state-based Trojans, Backdoors or Rootkits ? How does Linux respond to moral and ethical challenges ?
7.How can Linux foster major IT innovation - say a speech recognition hardware/software module that ties closely in with the X-Windows and interface systems ? Or maybe a BI/Stats engine that does very fast transactions but uses proprietary hardware+software with tight coupling to kernel required. This issue of proprietary and how it interfaces to Linux and Open Source is a frequently recurring issue - see here.
8.How does Linux resolve the quandary confronting all operating systems - how to retire and replace working and usable code packages with more efficient, performant and/or functional code. One only has to look at the number of shell command systems (command scripting languages in Windows and Mac Os) to see one example of the problems confronting not just Linux, but all OS suppliers. Customers have large investments in code dependent on the old code components, APIs, and scripts. They are reluctant to convert or abandon them - what do Linux and other OS vendors do to retire/deprecate such systems gracefully ?
9.How does Linux rate and evaluate its packages and components for reliability, performance and usability in a systematic and transparent/objective fashion ? How does Linux team use this information to determine the next version's features list and priorities ?
10.How does Linux provide for succession when Linus no longer wants to carry the leadership torch ?

The Overview

Looking at our Top Ten Facts about Linux, there are some immediate and obvious trends and conclusions. For the past ten years Linux, both on the desktop and the server, has been the most reliable, secure and IT plus Web standards compliant operating system. Linux provides on the desktop and with its servers a set of functional features and component architecture equal to if not better than their commercial counterparts such as Apple Mac OS or Microsoft Windows. There is not one current major OS trend in which Linux not only has kept technical pace but also is a major entrant and/or contributor - be that grid computing, virtualization, advanced security initiatives or SOAP and Web Services.

So it should not be a surprise that during the past ten years Linux has been the fastest growing operating system. But that raises the question - How could a free and open source operating system achieve such a record pitted against commercial powerhouses such as Apple Mac OS, Microsoft Windows, and various flavors of Unix ? The answer to that appears to be the fact that Linux is more suited to the reality that large swaths of operating system and general software infrastructure are becoming standard and commoditized.

Harvard Business School's Nicholas Carr raised the point at the recent Linux World conference. As major features of the OS, middleware, databases, and other software infrastructure mature and become more stable - Carr argued that they also standardize and as components become commodity priced or given away for free. Examples are Sun's open sourcing of Solaris while BEA, IBM, and Microsoft among others give away for free key components of their Application Server development stack. And in the database arena a free giveaway battle has broken out amongst IBM, Oracle, and Microsoft.

Microsoft started the ball rolling giving away for free starting in 2000 what is now an almost complete BI stack with its database. Oracle and IBM have responded in kind giving away free BI stack components and most recently complete Express versions of their databases. These Express versions incorporate the latest features with the exception of some high end enterprise capabilities. Users are allowed to use these copies not just for development work but also in production. And given that for example IBM DB2 Express allows up to 4GB and two dual core processors to work with DB2 Express, production installations of DB2 Express should be able to support scores if not hundreds of users. This is exemplary of the new free giveaway and bundling wars that have broken out in the software arena as technology starts to coalesce around key infrastructures: operating system, application servers, database system, mail and messaging systems; information collection, analysis and re-distribution; Web and system development. This consolidation has triggered a transition in IT business models in which Linux and Open Source are major players.

Live Free or Die

The whole idea that software can be bundled and given away for free is the result of the DOJ/Microsoft settlement of the Antitrust case in 2000/2001. The resulting verdict was so ambiguous and inconclusive on predatory pricing and bundling, that Microsoft and others have proceeded with the viewpoint that anything can be bundled for free with anything else. The clear examples are what Microsoft is doing now. There is the free BI Stack given away with SQL Server 2005 and the growing set of free utilities bundled with Windows XP and upcoming Vista.

What this means for start-ups and even existing pure play software vendors is that they cannot hope to charge for their software on a SaaL-Software as a License model for any period of time without making themselves vulnerable to attack. Even if a software vendor has superior features, performance and/or reliability, it may well lose to an entrance by a Microsoft, which can bundle for free their "just good enough" software program with database, application server, operating system or other infrastructure element and steal away market share from the smaller, pure play vendor. Over time a Microsoft then has the opportunity to improve their product so that it is better able to compete on its own. This is an oft repeated play and just what Microsoft is doing in the BI marketplace for example: giving away a free complete BI stack starting in 2000 with OLAP, ETL, DataMining, and now includes Reporting and Warehousing Services. In effect the monopoly players have enough monopoly profits a portion of which they use to "finance" their new software ventures. This is straight from the playbook written by Standard Oil's John D Rockefeller at the turn of the 19th-to-20th century. It was these predatory pricing abuses which resulted in the US antitrust laws in the first place.

So what can a start-up or smaller, pure play vendors do ? They can stay away from large markets identifying a niche and build out commanding shares there - become the big fish but in a small pond. Or ally themselves with one of the large players and either be bought by or act as as protected VAR/vassal/3rd party supplier for that large player. Or they can adopt the SaaS-Software as a Service model and offer their software as a service where the orientation is towards service level agreements plus Web and other modes of delivery which protect the smaller vendor. Or they can adopt freeware or an Open Source business model.

The Open Source model of which Linux is one of the prime examples, gives away not only the executable but the source code for free using that to attract both users and a broad developer community. Then Open Sourcers can, if they choose, make their money on service and making their software work for their clients. Linux Open Source players such as RedHat, Novell/Suse , Xandros, and Linspire can only prosper by innovating with features and functionality as fast as their rivals while making money on support, training and consulting services. This is a potent reversal of the SaaL-Software as a License model.

In the SaaL model as it currently stands software vendors say to their users - "here is what we give to you - take it or leave it ". And then they live or die based on how their well their wares are accepted in the marketplace. But dominant monopoly players like Microsoft (90%+ share in OS and Office desktop software, 80% share in Web Browser and system development tools, etc) are better able to say "we will determine what is in our software, at what time and with our priorities - which may or may not coincide with yours". The new Office for Vista is a prime example where major changes to the interface are being made like it or not. Ditto for the Windows Vista OS hardware requirements. This latter point is important when existing usable software becomes obsolete in the eyes of the software vendor and they wish to retire it on a schedule that meets their cost control needs and support efforts - not the users priorities and needs. Visual Basic is a classic example in that the transition to VB.NET left a whole development community either to wither away or be faced with difficult if not daunting conversion efforts. And yet VB.NET itself was seen by a sizable group of Visual Basic practitioners as abandoning key features of the language including ease of development.

Now it is true that all developers including Linux and Open Source are confronted with this obsolescence problem (see point 8 in our list of Top 10 Problems Facing Linux). However, the difference is EULA. In SaaL-Software as a License, by signing EULA agreements, users have ceded all legal rights to the software vendor. EULA agreements deliberately preclude any warranties of service, plus no performance, reliability nor other performance measures. This practice, again lead by Microsoft but with tacit co-operation from many other software vendors, severely limits the legal liability of their software in any disputes over non-performance or outright failure as we have seen with the continuing stream of scalability, reliability, security, and interoperability problems with Windows, Office and IE. In effect, the software vendors have transferred almost all of the risk of software failure or non-performance to IT buyers.

Linux and Open Source business model does not completely reverse this situation; however with Linux and open sourcers vying for the bulk of their revenues based on service, support, training and consulting work - there is inevitably greater commitment to their customers' success as evidenced in service level agreements and various other commitments (including contractual) for support. Also with the source code freely available, other parties can and often do fork or change the code base over in a new direction more compliant with customers needs and requirements.300 distributions of Linux are testimony to this forking trend.

The new forked versions of Open Source code can then set up their own service, support, and training revenue streams. This has been happening many time in the operating system, compiler languages, database and other open source software segments in the past 10-15 years. So the danger of forking plus the orientation towards support and service tends to force Open Source players to be much more attuned to customers needs than SaaL software vendors protected by EULA and/or monopoly power.

So the New Hampshire state motto, Live Free or Die, applies to software vendors - especially those trying to enter markets that are a)already have dominant or monopoly players b)and/or have commodity pricing as new features become more marginal and/or c)have the prospect of becoming big markets ($1B++) in the very near future. In all these cases, adopting a free software approach (but not necessarily open source) with emphasis on service, support, and training revenues, has the benefit of protecting the new vendor from attacks by monopoly vendors who are increasingly willing to resort to bundling, giveaways and other predatory pricing practices.

As a software buyer of Open Source, my entry costs have been reduced to near zero - and now I can differentiate on a vendor's service, support and the ability to deliver features. So I am inclined to consider the Open Source or SaaS startup and pure play software vendors a bit more than the EULA-shielded monopoly vendors despite their short term freebies(example- back in 1996 Bill Gates pledged that Microsoft IIS web server as well as IE browser would be perpetually free in his assault on Netscape. Full IIS can no longer be found on Microsoft's website for free download; and in fact forms one of the core modules in Windows XP Professional edition which costs $100-200 over and above Windows XP Home edition. So much for promises of "perpetually free").

This is the thrust of Harvard Professor Carr's and others remarks on the nature and maturity of core or infrastructure segments of software markets. As they become more mature, common or standardized in approach, they become more commoditized in pricing - so now vendors have to compete on training, support and service. Linux and Open Source vendors can well state that they and their communities are better oriented towards these goals.

Linux Prospects

Linux in 2006/2007 will be at a crossroads both on the desktop and as server. Linux in both markets has matched its major competitors for features and functionality for at least the past 3-5 years. But over the past 10 years it has has clearly outperformed it competitors in reliability, security and performance.

Microsoft Windows desktop and server has slowly (excruciatingly so for those who have had to endure the Blue Screens of death through rampant worms and virus attacks to the now monthly and often massive security updates) - has slowly narrowed (but not closed) the gap on reliability and security. But even with the new Vista, there are serious performance deficits relative to Linux systems - they can be identified as Windows Bloat, Windows Rot , and Windows Clot.

Windows Bloat means that Linux will run with same performance on a system with half of the hardware requirements for memory, CPU power and Video card support as Windows XP. This differential will only increase with new Windows Vista due out in early 2007 as Vista's basic hardware requirements move from 512MB to 1GB of memory, from 1GHz CPU to 2.4GHz, and from 32MB video card to 128MB. Windows Rot means that over time Windows performance on the desktop and some server installations degrades because of registry, swapfile and other system resource problems that have not been fixed in Windows Vista. The result is that in order to get the hardware back up to speed, the OS has to be wiped clean of the hard disk and re-installed. Windows Vista also does not apparently address this problem. Windows Clot means that if a desktop or server is exposed to too many viruses, malaware, and other hack attack detritus - the configuration, registry and other install/support files become so slow and clotted with hack debris that the only solution is again to wipe the hard drive clean and re-install the OS. Windows Vista apparently does address portions of this problem but the record on how effectively is yet to be determined.

These problems for Windows represent opportunity for Linux,Apple and other competitors who have generally been able to handle these problems. But perhaps the biggest opportunity being presented to Linux and other vendors is the growing dissatisfaction with Windows as development environ where only start ups and small pure play vendors can make money. We have already outlined in "Live Free or Die" section some of the problem confronting vendors when major monopolists decide to take over a market or use it as "free bait" for one of their complimentary products .Microsoft has been a major sinner here.

In effect Microsoft is increasingly taking market away from the very application vendors that have been the golden goose egg-layers for Microsoft for the past 20 years. Just ask the accounting and Small Business tool vendors. Or the remaining desktop project management tool vendors. Or the very small number of operating system utility vendors remaining of what used to be a vast coral reef of Windows support facilities. More recently Security vendors and BI players have had to adjust to the fact that Microsoft would not only be entering their markets but also offering giveaways and free bundles to accelerate Microsoft uptake at the expense of their "partners". This is the opportunity for Linux vendors to differentiate themselves from Windows and Microsoft..

Unfortunately, Linux vendors may not be prepared to take advantage of the huge opening presented by Microsoft. The problem is that many Open Source founders look in askance at SaaL and commercial software. Now this party believes that SaaL+restrictive EULA is worthy of their concern if not out right scorn. But many commercial software vendors use various forms of EULA in a much less restrictive fashion than say the Microsoft model. Other commercial vendors are moving towards service level agreements in their contracts with customers. In general, software vendors are having to respond to the sometimes predatory practices of their dominant infrastructure providers with diverse but more service and support oriented management.

But moving to Linux is not a welcome proposition if a commercial vendor is going to be greeted by free software participants as a commercial pariah. For example, in my reviews of three Linux distributions I found very few examples of freeware and no examples of commercial limited duration trialware available on the 3 distributions examined to date. In the Linux magazines and websites there is a dearth of coverage of even important new emerging commercial Linux software like RealBasic or XaraLinux or DB2 or Oracle. At the LinuxWorld conference I talked with three commercial software vendors who had failed to get their trialware even partial distribution on either RedHat, Suse, or Xandros Linux. There is also uneven coverage of the Linux distributions and what target segments they are trying to address. In sum, this is the number one problem confronting Linux vendors - how to accommodate commercial, SaaL vendors.

Linux vendors may be driving away the golden goose - applications, applications, applications that brings users, users, users. And in fact, SaaS vendors who develop primarily for Web Service or desktop Web delivery have been getting the lions share of venture capital in the last few quarters. How and which distributions adopt a more open, commercial-software-receptive stance is still to be determined. But when Oracle's Larry Ellison recently indicated an inclination to buy and sponsor an Oracle-Linux distribution, user reaction was very positive. This is a very telling indicator of the direction where a broad set of users think the Linux market should be heading.

Many of the remaining problems we cite are variations on this theme ( see problem points 5, 7, 8 above). How does Linux resolve its free and open roots with vendors whose business model is different and commercially oriented - software should be paid for upfront or on an ongoing basis commensurate with benefits received. The long term success of Linux will be determined by how well it becomes non-secular, able to embrace developers and users of different faiths or business models into its fold.

The other major problem confronting Linux which Apple and Microsoft will hammer home at every opportunity is the wide range of distributions with disparate install, system management, shell scripting and GUI interfaces or programs. This makes for a learning nightmare and provides the only live and real ammunition in Microsoft's Get The Facts Campaign. Linux distributors and supporters appear to be addressing some of these issues. There is talk of a common interface spec and a set of standard APIs for the KDE and Gnome GUI interfaces. Ditto for system management and install procedures although the latter are in the very basic infancy stages. Finally, there are the problems of succession. Both in leadership and in replacing old components and systems both Linux and Microsoft face common problems - what comes after Linus Torvalds and Bill Gates ? Who or by what decision making process will old but still functioning components be retired from Linux or Windows ? How graceful and with what support for the existing user base will these "retirements and replacements" be done. This has been variously done by scorched earth through to laissez faire policies and is a problem faced by many software infrastructure vendors.

Summary

Linux and Open Source software are at a crossroads. Microsoft seems committed to a version of Windows in Vista that will give Linux and other OS software a distinct competitive advantage, particularly in the desktop market. Also Linux and Open Source are better adapted to the service orientation that will determine success in the increasingly commoditized and standardized software infrastructure markets. But Linux is confronted with challenges of its own. The Linux distributions and community will have to be welcoming of other business models represented by commercial software distributions. In the long haul, Linux success will be measured in how many useful applications migrate to the platforms. Microsoft's Steve Ballmer's almost ironic cry for more developers, developers, developers on Windows could be translated for Linux into " more applications applications applications" because that means "users users users". It will be interesting to see how this situation unfolds over the next two years - we promise more coverage.