I would stop well short of making a blanket statement that open source development automatically ensures technology innovation.

In the majority of cases, open source software projects have done an excellent job of reproducing what has already been created by software development companies. We believe there are a number of factors that need to exist for such a project to grow on its own. Those factors include a market where proprietary products exist that have well-defined the functionality – that is needed to compete in the market. There needs to be broad expertise among the developer community about how to build such products, and most importantly, there has to be motivation for a project to get underway and sustain its momentum.

Using Linux as an example, Linux has since its inception been designed as a clone of the Unix operating system. Today, Linux offers most of the same capabilities as Unix does, although it still lacks some of the features applicable for (and required by) customers that use the largest and most critical Unix installations. The Linux operating system is not an example of technology innovation, it is an example of how open source development can reproduce existing, known functionality for multiple architectures in open source form … carrying with it the use rights ensured by the GPL.

Along the same lines of open source technology innovation, in some cases, open source projects can be lacking in long-term vision. It can be challenging to get a technology road map that extends beyond the current project goals. On the other hand, an interesting approach we see today with commercial organizations leading a “community” project, where the commercial organization spearheads the technology roadmap, and taps the community for development support, and shares back the result of the community effort.

Al Gillen
Research Director, System Software
IDC

Left by agillen on October 3rd, 2005

Frank, Al,

How can you claim that Negroponte is wrong on Linux market share when you count only boxes and don’t account for multi-OS servers?

Left by ARonaut on October 6th, 2005

Response to ARonaut’s comment/question:

Servers supporting multiple operating systems are growing in presence today, but the actual number of servers supporting multiple operating systems is still low (as in single digits of market share). Going forward, servers supporting multiple operating systems will become more significant, but remember that the rising tide of virtual machine- and partition-supported operating systems will lift boats other than Linux… this trend will also float Windows installed bases higher. As a result, there won’t be a single OS that benefits dramatically better than other competitors due to this trend.

Taking this discussion a step further, remember that in many cases, the use of multiple-OS environments does not always result in net new OS deployments coming into use; instead, some of these OSes will be consolidated from other platforms, in effect, extending the life expectancy of OSes without necessarily uplifting the total number in use on a 1:1 ratio. We believe there is a fair amount of Linux that was installed on very low end hardware (in some cases, aboard PCs or workstations that are configured to function as a server), and many of these installations are candidates for consolidation or rehosting.

Finally, IDC’s current OS shipment and installed base data for paid OS instances considers OS deployment without regard for whether it’s deployed on a dedicated or shared machine. The market share for the installed base of Linux server operating environments was 12% in 2003, growing to 14% in 2004. It is important to note that the installed base percentage lags new installations percentages since the installed base is an accumulation of shipments over the past 5 years. (Please see IDC document #32452 for a much longer discussion on this topic.)

Hope this helps clarify.
ag

Left by agillen on October 6th, 2005

I’d add to Al’s comments:

There are a lot of different ways to gauge Linux "presence" and "impact" in the market. I first laid out server shipments with Linux loaded as one proxy. Al Gillen contributed the installed base figure (above), which gives yet a lower figure. My guess is that an estimate of the percentage of production workloads running on Linux would be lower still.

But wait… is this reality somehow "bad news" for Linux’s future? Certainly not. Linux is still in early days, relative to the other major OSes – yet it’s been adopted by some pretty big organizations as one of several key platforms. It has a very strong growth trajectory (as seen in IDC’s server shipment figures). And – most importantly for its future – it has a fast-growing community of developers and integrators focusing on it as a platform.

So, I’d say "don’t cry for Linux, Argentina" (apologies to Tim Rice, Andrew Lloyd Webber and Argentina). On the other hand, let’s be realistic about where it is today.

Frank Gens
IDC

Left by fgens on October 6th, 2005

[...] I shared Nicholas Negroponte’s comments about open source at IDC’s European IT Forum. His assertion is that the open source model is [...]

Left by IDC eXchange » Blog Archive » Customers Think Open Source Model Will Drive Better Software on February 17th, 2009