Carl Weinschenk spoke with David A. Wheeler, widely known open source analyst.
Weinschenk: How do you determine total cost of ownership in open source?
Wheeler: Determining TCO is simple, in principle. Just add up the estimated costs over the time of use. Some costs, such as training, typically are similar for open source software - OSS - and proprietary software. The initial licensing fees are often different. OSS often has little or no initial licensing fees, while the initial licensing fees are often substantial for proprietary software. This means that you can often "try out" OSS as long as you like, and then pay for support later once the software becomes more important to your business.
Weinschenk: So does a company need to invest in parallel development if it is "trying out" the software? What if it doesn't measure up?
Wheeler: It depends on what the company is planning to do with it. If you must have a particular function, then clearly trying out multiple different OSS options or not is part of the process of making a rational decision. Different organizations do it differently based on the level of risk.
Weinschenk: What happens after the purchase?
Wheeler: Support is probably the most remarkable difference between OSS and proprietary software. If you're basing your business on the software, you need to fund some kind of support plan, whether it's OSS or proprietary. For proprietary software, you really have only one support option - the original vendor - because they are the only ones who can actually fix important defects.
Weinschenk: In a proprietary situation, can support come from an unconnected third party?
Wheeler: Other parties can act as support for proprietary software, but in some sense they are only middlemen for the original software vendors. That's because they usually cannot fix defects in the software or make substantial changes to it.
Weinschenk: Does this affect TCO?
Wheeler: Sure, because TCO includes the product supply function. There can be errors that are accepted and the company moves on. Some are more serious and must be corrected or the company will suffer major losses. You may say that in the proprietary world the company is dependent on the maker of the software to correct things. If it is a very large fee to make it so, you are going to have to pay that large fee. That is one of the major differences between OSS and proprietary software.
Weinschenk: The sheer number of OSS options available must be good for TCO.
Wheeler: That competition often drives down prices. For popular OSSes, there are typically many options, such as companies that specialize in supporting that particular software. Red Hat, Novell and Sun are examples. There also are more general support organizations. A remarkable number of organizations do OSS support internally. But for most organizations, I suggest paying for the external expertise just as you would for proprietary software.
Weinschenk: Why do you suggest going outside the organization?
Wheeler: Fundamentally, a lot of organizations do not have the internal expertise, nor do they need to. You use outsiders for the same reason you pay specialists to do other things.
Weinschenk: What other costs are there?
Wheeler: Transition costs are something else to consider. In the proprietary world, there often is significant cost to change to anything else, be it proprietary or OSS. This is no accident. Proprietary software developers have a large financial incentive to make this so. But transition costs are not necessarily a reason to stay with the same software. Vendors often have high annual support costs. Consider how long you'll be using the software, including its later versions. Once you consider the total costs over time, the transition costs may very well be worth it.
Weinschenk: Are there transition costs with OSS?
Wheeler: There often don't tend to be large ones. The reason is pretty simple: With proprietary software, the vendors make it difficult to transition by hiding how the data is stored. It is fundamentally not really possible to hide how it is stored in an OSS product. It is not really possible to hide the open source programs. As a result, there may be some costs, but they tend to be small because the key problem - lack of information about stored data - isn't an issue.
Weinschenk: You suggest that improvements made by a company in OSS software should be resubmitted to the main project. How does this relate to TCO?
Wheeler: If you don't submit the improvement back to the project, the version you have will increasingly be out of sync with the main body of work. You will lose the benefit of that main body of work as the inconsistencies between what you have and that main body grow. TCO will grow because you will be maintaining your original version, and that will be more troublesome and expensive. Most of the cost is in maintenance. You want to share costs with as large a group of people as possible.
Weinschenk: What is the biggest challenge in making people take advantage of OSS and the TCO benefits it seems to offer?
Wheeler: Probably the biggest problem today is that many people don't realize that OSS is commercial software, and thus never even consider it as an option. A lot of people say, "I only want commercial software" and mistakenly ignore OSS. They don't realize it is commercial software. The whole reason to use commercial software is because it creates an economic model where multiple users can share resources to develop better products.