Linux Saves Money and the Numbers Prove It
A few days ago, Cybersource, the company I work for, released a study undertaken on the relative operating costs of running a mid-sized business on Linux and free software on the one hand, and Windows and Microsoft applications on the other. We've had a significant response to this study from many industry news organisations, industry players and individuals. What I'd like to discuss here is both the context within which this study originated and the methodology and results which precipitated.
First, a little history. While our firm has been using Linux for almost the entirety of its ten-year history, (preceded by a stint with Minix no less), we have been recommending Linux server and network infrastructure solutions for only about six years and line-of-business desktop solutions for about three years. In all circumstances, we want to know that Linux could solve the business problems of a client before we proffer a solution recommendation. The Linux server solutions business has been going strongly, but uptake of Linux desktops among business customers has been slower. The past nine months, however, has seen a dramatic increase in the number of existing and potential new customers, indicating that they are looking seriously at large-scale, non-infrastructural deployments of Linux desktops and terminals. Many of these customers approached us with the question: "How much will we save if we move groups, departments or the whole company to Linux?" We didn't have an answer for them then. We do now.
One of the other prime motivators in our undertaking this project was as a response to Microsoft's claims that the cost of software licensing was a virtual non-entity in the overall calculation of IT budgetary expenditure, accounting for a mere few percent. Our experience with many clients indicated otherwise. We began with our first study, released late last year.
This study looked at the differences in software license costs between Linux and Windows, for 50, 100 and 250-user organisations. We included the types of core applications that most organisations need: workstation operating systems, office productivity, network infrastructure servers, file, e-mail and database servers as well as intranet, internet and e-commerce systems. As you can imagine, Linux easily defeated Microsoft's platforms and applications in a license cost comparison. This study was well received, but some of our readership recommended we look beyond just software license costs. We agreed. To gain the attention of financial and non-IT managers, we needed to provide a holistic 'real world' model, encompassing hardware, networking, staff, consultants, internet fees and sundry other cost components. Thus, this new study provides such a comparison.
We opted to base the costs on the IT needs of a 250-user organisation. We believe this would provide a reasonable indicator and gnomon from which to calculate IT costings for larger enterprises, as well as smaller firms. We decided to run two models, representing organisations that are buying everything from scratch and organisations who will use their existing hardware and cabling. For our green-fields model, we specified the servers based on our experience with equivalently sized customers' sites. We set the per-node network connection cost (inclusive of all cabling, switches and hubs) to a set figure and specified the workstation hardware on current generation business-class systems from a name-brand vendor that supports both Linux and Windows. We applied this same total hardware and cabling cost to both the Linux and Windows platform models.
For all the prices upon which we based our final calculations, we provide full details for the configuration of the hardware, a vendor-sourced URL, as well as a method by which we arrived at costs. We wanted to make sure that our own results could be verified or repudiated, thus enhancing the study. This theme ran throughout our work in general. We avoided all instances of cost incursion for factors that we could not provide complete verification for or that were at all disputable. We believed this would tarnish the overall results and weaken the paper's impact. We also wanted to run the model over the industry-accepted norm of a three-year IT equipment purchase/deploy/use and decommission cycle. This will hopefully allow the vast majority of enterprises whose own processes follow this cycle to apply the study's results directly to their IT-budget spreadsheet projections.
Other cost factors included in our calculations are staff wages, consultants' fees, costs for line-of-business software, the purchasing of specific technical applications, internet connectivity and a miscellaneous costs category as a final "catch-all". The figures we arrived at (based on re-using existing equipment) were $733,973 US for a Windows/Microsoft solution and $482,580 US when using Linux and free software, a saving of $251,393 US or 34.26%.
When the model requiring totally new hardware was computed, the Microsoft-platform solution cost $1,042,110 US, and Linux came in at $790,717 US. This implies a 24.69% reduction. It also shows that outfitting and running an organisation with brand new servers, networking, workstations and choosing Linux software is comparable in cost to keeping all the old equipment and purchasing a Windows platform solution. In view of the forthcoming licensing changes that Microsoft will soon bestow upon its customer base, whereby most organisations will need to repurchase all their existing Windows systems and applications at non-upgrade retail prices, this scenario will become very common.
The other interesting factor that came out in our research, is that IT professionals who have solid Linux skills appear to be paid a premium over correspondingly attired Windows-platform specialists--obviously good news for those who are presently on track towards Linux Professional Institute or RHCE accreditation or who have a strong leaning toward professional Linux and free software services. It must be remembered, however, that even with this cost disparity, the organisaton that opts for a Linux solution will manage to slash instantiation and operating costs severely.
In fact, it is this final point that now needs to be expounded. If, as the study has shown, many, perhaps most, organisations can accrue substantial financial benefits through the partial or total adoption of Linux and open source software, what remains to convince this group to move wholesale to Linux? Are existing implementors of IT within these organisations the remaining roadblock? Are they concerned about, as Australian art historian Robert Hughes famously pronounced during a corresponding period within the art world, "the shock of the new"? Would faster progress be made if these results were presented to financial and non-IT executive company officials, in the language and format they know and understand, rather than an IT staff? Would many of the businesses worldwide presently suffering economic hardship (and shedding staff) benefit financially from such a move to Linux? I welcome your responses.
Con Zymaris is CEO of Cybersource Pty. Ltd., a long-standing IT and internet professional services company.
Getting Started with DevOps - Including New Data on IT Performance from Puppet Labs 2015 State of DevOps Report
August 27, 2015
12:00 PM CDT
DevOps represents a profound change from the way most IT departments have traditionally worked: from siloed teams and high-anxiety releases to everyone collaborating on uneventful and more frequent releases of higher-quality code. It doesn't matter how large or small an organization is, or even whether it's historically slow moving or risk averse — there are ways to adopt DevOps sanely, and get measurable results in just weeks.
Free to Linux Journal readers.Register Now!
- August 2015 Issue of Linux Journal: Programming
- Hacking a Safe with Bash
- Django Models and Migrations
- Secure Server Deployments in Hostile Territory, Part II
- The Controversy Behind Canonical's Intellectual Property Policy
- Huge Package Overhaul for Debian and Ubuntu
- Shashlik - a Tasty New Android Simulator
- Readers' Choice Awards 2013
- General Relativity in Python
- diff -u: What's New in Kernel Development