An Australian IT services firm has updated what it says is one of the few fully transparent studies comparing the costs of running Linux vs. Windows - and found that Linux is still cheaper.
The new study [pdf], from Melbourne-based CyberSource, found that Linux installations can be up to 36 percent cheaper to install and run over a period of three years than comparable Windows systems, though subscribing to enterprise technical support and buying new hardware and infrastructure can lower the savings to as little as 19 percent.
The study was the first Linux/Windows TCO (total cost of ownership) study when it was originally released in 2002, claims CyberSource. It has now been updated to reflect cost changes since then, such as the introduction of mandatory support contracts by some Linux providers and Microsoft's controversial Software Assurance subscription policy.
The main reason why companies should take notice of the study is its transparency, the company claims. While a number of TCO studies have appeared in the past two years, surprisingly few are fully open about the methodologies they use to calculate cost. Others, such as two controversial studies from The Yankee Group and IDC, have been shown to be biased.
CyberSource based its calculations - what it calls a "first-pass quantitative estimate" - on the average computer-usage requirements for an organisation with 250 users over a three-year period. The costing models include expenses such as workstations, servers, networking, IT staff, consultancy fees, Internet service charges, file, mail and print servers, e-commerce servers, SQL and network infrastructure servers, Internet and LAN servers, line-of-business software, desktop productivity applications, external training, printers as well as miscellaneous systems costs.
The study looks at re-using existing hardware or buying new hardware and infrastructure, and also contrasts Linux installations without third-party support with those with mandatory support contracts, such as Red Hat Enterprise Linux. The idea is that a company of any size should be able to scale the figures to apply to its business, CyberSource said.
The lowest-cost system was a Linux system without external support, using existing hardware, which cost 36 percent less than a Microsoft system on existing hardware. Linux without external support but on new hardware cost 26 percent less than Windows. A Red Hat system with a support contract on existing hardware cost 27 percent less than Windows, and Red Hat on new hardware cost 19 percent less than Windows.
CyberSource said that since the company is known as a Linux systems provider, it took care to stack the deck in favour of Microsoft in several ways, such as factoring out the cost of malware attacks and tripling the cost of external consultants for Linux installations.
"We've given Microsoft every head-start possible but Linux's cost advantage is simply too great for most organisations to ignore," said Cybersource chief executive Con Zymaris in a statement.
Why should CyberSource's study merit any attention amongst the propaganda coming from both sides of the open source/proprietary debate? The company argues that this is one of the few places a company can find a fully open analysis of Linux and Windows installations. "We explain, in sometimes droning detail, each and every step of our analysis," the study says. "We also provide sources to every stipulated line item cost, for both Linux/open source and Microsoft Windows platforms. To our knowledge, no other TCO model sampling does this."
This is in contrast to several other major TCO studies, CyberSource says. A 2002 IDC study called "Windows 2000 Versus Linux in Enterprise Computing", for example, found Linux was more expensive than Windows. But this was funded by Microsoft, and more importantly, one of the report's authors later said Microsoft had chosen scenarios for analysis that would be more costly using Linux, CyberSource pointed out. The study is no longer promoted by Microsoft.
An April 2004 report from Yankee Group called "Linux, Unix and Windows TCO Report, Part 1" surveyed 1,000 IT managers across various types of organisations and found that most believed Windows offered better TCO. But CyberSource noted that it was later made clear that the sample group was taken from a mailing list aimed at Windows system administrators.
A Meta Group white paper from May 2004 called "Linux Servers: No Silver Bullet for Total Cost of Ownership" found that higher costs for applications and support raised Linux TCO to levels comparable to Windows. CyberSource said it factored such costs into its study.
An April report from Forrester called "The Costs and Risks of Open Source" found that operating expenses for Linux can be higher than for Windows "for some companies". The survey sample is not exhaustive, Forrester said; while some of its findings were based on a survey of 140 North American companies and in-depth discussions with 14 of those companies, only five of the companies had kept detailed metrics on TCO. Those five companies said their Linux costs were five to 20 percent higher than their current Windows installations, and two of those companies said they expected their costs to go down as they gained experience.
The only major independent study to contrast Linux against Microsoft is a report from Germany's Soreon Research, using data collected from interviews with 50 enterprises, CyberSource said. The report found that Linux had up to 30 percent lower TCO than Windows.
CyberSource's report can be found here [pdf].