In its comparison of Windows and its .Net development tools against Linux and J2EE-based tools, Forrester found that large companies developing a custom Web-based application using Windows will save 28.2% to create the application, then maintain and support it for three years. A midsize business should save approximately 25% if it relies on Windows and .Net rather than Linux and J2EE as its development and support environment, said John Rymer, a Forrester VP and one of the analysts who authored the report.
The study was done by interviewing IT managers at a dozen organizations-- seven that use Windows, five that use J2EE on Linux platforms.
The small size of the sample may give Linux supporters reason to question the findings, but Rymer defended his methodologies. "It's like reporting," he said. "After a while, you start hearing the same thing. I think we got a good average idea" of costs and development time.
The main drivers of the big differences in development costs, Rymer said, were due to higher licensing costs of J2EE and databases sitting atop Linux, and the longer time necessary to develop the proposed applications in the J2EE/Linux environment.
In the big-company scenario, the licensing costs for J2EE tipped the financial scales at $208,567 for BEA application servers, Oracle databases, and BEA development tools. Costs for the equivalent Microsoft licenses, however, would run only $52,591. Advantage Microsoft, to the tune of $155,976. That difference accounts for about a quarter of the cost savings Forrester claimed companies would see by going with the Microsoft's platform.
The other factor is time: It takes businesses longer to develop and deploy custom Web-based applications using J2EE/Linux--on average three months longer, the report said. That translates into labor costs savings for Microsoft shops of $331,000, or about half of the .Net advantage.
But Rymer cautions against taking his findings and making general degrees about Windows vs. Linux. "Frankly, we're tired of these kind of broad statements," he said. "In the real world, there are different scenarios and skill bases and cultures. This only focuses on one scenario, admittedly a broadly-used one, but you can't use it to generalize."
Gartner analyst Michael Silver agrees. Silver co-authored a report released Tuesday by Gartner on the return on investment companies may see by migrating from Windows to Linux on the desktop.
In his narrow-cast look at Windows vs. Linux, Silver found advantages to both camps. "There are situations in which a move to Linux on the desktop will deliver ROI and does make sense," he said. On the other hand, those scenarios are quite narrow in scope. "Migrating desktops to Linux only makes sense in a very limited range of situations," he added.
Although Gartner's analysis showed that companies shifting to Linux can save $80 on hardware acquisition costs and an average of $74 per user per year on office productivity software compared with Microsoft's offerings, he noted that those companies could realize the same cost benefit simply by shifting from Microsoft Office to the Windows editions of Sun's StarOffice or the OpenSource.org suite, which is free.
"Enterprises need to separate the operating system decision and the office [productivity applications] decision," he said, noting that companies can move away from Microsoft Office and see some cost benefits but keep Windows for the hundreds of other applications they use.
In some instances, Silver said, Linux might be a good call. "Linux may merit a migration," he said, in situations where companies are thinking about moving employees who use a limited number of applications, such as data entry, call center, and other structured-task workers. And companies still relying on aging versions of Windows--such as Windows 95 and 98, which are unsupported and barely supported, respectively, by Microsoft--will likely see a bigger return by shifting to Linux than companies that use the newer, and more reliable, Windows 2000 and Windows XP.
Silver noted that while Linux may have a lower up-front cost, thanks to lighter hardware demands and a lower-priced client operating system, that's not the whole story. "Other costs, such as labor, training, and external services must be considered," he said.
And there, Linux may not meet the needs of corporations, or give a good return on investment. "Linux support on the PC is the exception and not the rule, a challenge that an enterprise must consider when determining the future of its desktop OS."