Over the last several weeks, a debate has been raging between Windows and Linux proponents over the total cost of ownership (TCO) of their respective operating systems. At the heart of that debate is a new study by International Data Corporation, which claims that for most workloads, Windows 2000 could have a lower TCO than Linux.
The research team at IDC came to its conclusions after interviewing IT managers at 104 North American firms about their costs for acquiring and supporting Windows 2000 and Linux servers. The team studied costs for five different workloads: network management, file serving, print serving, Web serving, and security. For each workload, IDC looked at costs in six areas: hardware, software, IT staff, downtime, training, and "outsourced" costs for assistance from external service providers. After standardizing the cost information from all firms, the research team applied its data to a hypothetical "average" company with 1,315 users to determine total costs over five years. The following table shows those costs on a "per 100 user" basis; this allows firms of various sizes to scale the costs to fit their own situation.
Windows 2000 and Linux Five-Year Cost of Ownership by Workload
(Costs Are per 100 Users) |
|||
Workload Type
|
Windows 2000
|
Linux
|
% Advantage
|
Network management
File serving
Print serving
Web serving
Security
|
$11,787
$99,048
$86,849
$32,305
$70,495
|
$13,263
$114,381
$106,989
$30,600
$90,975
|
Windows--11%
Windows--13%
Windows--19%
Linux--5%
Windows--23%
|
It is important to note that in all of the workloads that IDC studied, IT staff costs stood out as the biggest cost component, representing an average of more than 62% of total costs. This finding had a critical impact on its TCO estimates, as most of the cost advantage that IDC gave to Windows 2000 came as a result of its lower IT staff costs. Indeed, if IDC had decided that staff costs for administering Linux were essentially the same as those for administering Windows 2000, Linux would have had a lower TCO than Windows 2000 for both Web and file serving, and would have finished in a dead heat for security workloads.
A Closer Look at the IDC Study
The IDC study reflects Microsoft's assertion that while Linux servers may initially cost less than Windows servers, their advantage washes out over time because Windows administrators are more abundant and cheaper. Microsoft has a point here, as most salary studies have shown that Linux administrators do cost more. For instance, a survey conducted by PayScale places the median salary for individuals with Linux skills at $54,000 compared to $50,000 for those with Windows 2000 skills. In another study, Robert Frances Group found that companies paid their Linux administrators an average of $71,400 versus $68,500 per Windows administrator.
At the same time, there is evidence that the average Linux administrator manages significantly more Intel servers than a Windows administrator, a fact that could make them more cost-effective. In the Robert Frances Group study, for instance, Linux administrators reported that they managed an average of 44 servers compared to only 10 servers for Windows administrators. This big difference is probably due in part to differences in the kinds of servers these two groups manage. Linux servers are more likely to be low-maintenance "appliance" devices such as firewalls and packet filters; Windows, by contrast, tends to dominate the high-maintenance workloads such as file/print and messaging servers.
Still, anecdotal evidence from employers and IT recruiting firms indicates that, all things being equal, Linux administrators can often manage more servers than Windows administrators. One reason they can do so is the greater reliability and reduced downtime of Linux servers, a fact that the IDC study acknowledges. In addition, Linux tends to impose less paperwork on IT staffs. For instance, while Windows administrators must conduct regular audits of license usage and file their reports with Microsoft, Linux administrators are virtually free of this burden.
Windows Versus Linux: The Bottom Line
With all the claims and counterclaims being made about Windows and Linux costs, it can be hard to determine which operating system will be most cost-effective for your organization and your particular workloads. As an analyst who has studied both environments, I would offer the following guidelines for making the right decision.
- Consider your workloads. If you intend to run an application on Linux that is essentially a "load it and forget it" workload, there's a good chance that you will save money versus running that workload on Windows. However, if a Linux server takes more than two or three hours per week of an administrator's time, check out the Windows alternatives to see if they require less effort to manage.
- Consider your IT staff's skills. If your staff has no experience with Unix or Linux but plenty of Windows administration skills, the cost of hiring a Linux administrator or relying on external service providers for help could cancel any TCO benefits you might achieve. Check out the going rates for Linux administrators in your area before making up your mind.
- Consider what software standards your applications support. If your applications depend on Microsoft programming standards and systems management tools, the cost of migrating them to Linux will cancel any TCO benefits you are seeking. The only places where Linux will make sense for your firm will be at the edges of your network in "appliance" workloads such as proxy serving and intrusion detection. However, if your applications use Java and Unix standards or work with systems management tools that are available on Linux, chances are good that you can run those applications at the lowest TCO in an open-source environment.
In short, chances are good that for any given workload, whatever operating system your IT staff is already using (with the understanding that Unix and Linux are the same OS) and writing applications to will be your most cost-effective platform. That may be bad news for the Windows and Linux zealots who want you to win you away from their archrival's camp, but it is generally the truth. Moreover, it's a truth that all the TCO studies in the world will likely never change.
Lee Kroon is a Senior Industry Analyst for Andrews Consulting Group, a firm that helps mid-sized companies manage business transformation through technology. You can reach him at
LATEST COMMENTS
MC Press Online