Getting Down to the Business of Open Source

Readers will doubtless be relieved to learn that I do not intend writing one of those tiresome “top ten predictions for open source in 2011” - not least because I am firmly of the persuasion that those who live by the crystal ball are...


Readers will doubtless be relieved to learn that I do not intend writing one of those tiresome "top ten predictions for open source in 2011" - not least because I am firmly of the persuasion that those who live by the crystal ball are condemned to eat broken glass.

But I must admit that it would be nice to have something that steps back and looks at the bigger picture of open source in business, and happily the free online journal "Open Source Business Resource" has obliged with an issue entitled "The Business of Open Source".

The first article, and probably the one that will be most useful to readers of this blog, is entitled: Cost Optimization Through Open Source Software. As the abstract explains:

iXsystems wanted to quantify the benefits associated with the use of open source software at their company headquarters. This article is the outgrowth of our internal analysis of using open source software instead of commercial software in all aspects of company operations.

As you might expect, many of the savings produced by using free software came early on:

these costs add up to a grand total of over $2 million in initial software licensing and support package costs. This is in stark contrast to the total payout of $500 for all of the open source applications iXsystems uses in its daily operations. This allows for a significant reduction in the operational bottom line, which is certainly desirable for any organization, especially those just getting off the ground.

So far, so conventional. More interesting are the following thoughts on running costs:

At first glance, systems administration costs seem to favour the hiring of Windows administrators. In Silicon Valley, where iXsystems' base of operations is located, the median salary for a Unix administrator is $104,000 versus $85,000 for a Windows systems administrator. However, Unix system administrators are generally highly experienced and have a diverse skill set. A competent Unix administrator can support upwards of hundreds of desktop, server, and database systems. This is partly a result of the integrated design of Unix operating systems; once the base system is configured, the administrator can script many maintenance and administrative tasks. iXsystems manages to run a mail server, over 40 desktops, around 20 laptops, over 30 production servers, as well as many test servers with one experienced Unix administrator.

That's a good point, as is the following analysis:

In a Windows environment, specific services are provided by different products, with each product requiring a discrete skill set. Due to the complexity of configuring and maintaining these products, it is rare outside of a very small network to find one administrator who is skilled in all of the products required in a typical business environment, for example, MS Exchange, SQL server, IIS, Sharepoint, domain controllers, print servers, and desktop support.

Due to these considerations, iXsystems determined that, if it were to run in a proprietary environment, it would need a dedicated database administrator, an MS Exchange administrator, a desktop support technician, and up to two general systems administrators. This would bring administrative support costs closer to $400,000 for proprietary software versus a little over $100,000 for open source. This gives a cost advantage of around $300,000 per year to iXsystems for administrative support, freeing up further valuable resources for the core aspects of the business and lowering the bottom line.

I've not seen this issue explored before, and I think it's a subtle but crucial point that needs to be highlighted more in discussions about the "real" cost of both free software and proprietary solutions.

As was the case for the initial outlay, savings thanks to widely differing upgrade costs are less of a surprise:

For most enterprise applications, a company will upgrade their software at least every five years, and some applications will have even shorter upgrade cycles. In our example, proprietary software would carry a cost of $1.3 million every five years for software upgrades. With F/LOSS, the cost of upgrading software to the most recent versions is negligible and can be considered part of the salary costs of the Unix system administrator.

The following point is worth noting:

Using open source also carries the hidden advantage of being able to keep up with the latest improvements in software as they are released, rather than waiting until budgetary constraints allow. The end result is increased and improved application features sooner rather than later, without the increased costs.

The final tally is pretty much what you'd expect, but impressive nonetheless:

With F/LOSS, the costs over the course of five years are around $733,750 for a company with a setup comparable to iXsystems' current operations. If a similar company wanted to opt for proprietary software, they would have to spend approximately $5.2 million over the same time period. That is five times the cost for the first five years of operation. Over the course of 10 years, a company comparable to iXsystems could save over $8.7 million in software-related costs.

The second article, Best Practices in Multi-Vendor Open Source Communities, comes from Ian Skerrett, Director of Marketing at the Eclipse Foundation, who naturally underlines the benefits of the multi-vendor approach taken by his organisation. He makes the following important observation:

Successful multi-vendor open source communities do not require copyright assignments to a single for-profit company. This is because organizations and individuals want to participate as equals in a community. If one entity is aggregating a special right, in this case copyright to the code, it creates a two-tiered system. Copyright to contributions should be retained by the originating contributor or a license should be granted to a not-for-profit foundation. For example, copyright of contributions to the Linux kernel or projects at the Eclipse Foundation remain with the contributor.

Copyright assignment can be a real bone of contention among free software developers, who naturally dislike having to give up their rights to a company. As Skerrett points out, multi-vendor organisations largely avoid this problem because no member wants any other member to have a particular advantage, and so opt for no assignment. He also rightly notes that multi-vendor foundations are an increasingly popular way of organising communities:

open source foundations have demonstrated a scalable model for creating multi-vendor communities. Organizations such as the Apache Software Foundation, Eclipse Foundation, Linux Foundation and many others have implemented many of these best practices for their communities. Starting a new open source project under the auspice of one of these foundations allows for a project to start in less time and at a lower cost than without the help of the foundation, plus it also offers the benefit of leveraging an existing community.

Regular readers will know that I, too, am a big fan of foundations, so I definitely concur with Skerrett's analysis here.

The other contributions to this issue are also worth reading, especially How Firms Relate to Open Source Communities, which offers a formal analysis of the relationships of businesses and free software.

All-in-all, "The Business of Open Source" offers some useful stuff to get the mental wheels turning after the holidays at the start of what will doubtless be another exciting and successful year for open source in business.

Follow me @glynmoody on Twitter or

Find your next job with computerworld UK jobs