Business Models

In a recent comment, Savio differs with my opinion on what customers are purchasing. I believe that companies don’t buy open-source products. Rather, I believe, they buy the support services and infrastructure that goes with those products. Savio believes that customers “…want a product, and support is one aspect of a product offering, not the product offering itself.”

This may be a matter of semantics but I think the differences are important. I’ve argued that the proprietary licensing model for software was an important economic innovation when it was first implemented. I’ve also argued that the proprietary licensing model is now obsolete and the open-source model is a fete accompli that cannot be stopped.

When companies like Oracle and Microsoft first started they had woefully incomplete products in a market where database and operating system skill sets were very rare. Most people had not even heard of SQL let alone possess the capacity to understand SQL parsing, optimization or execution. Oracle had a business model where they would collect a relatively small fee from a relatively small number of companies in order to fund the basic R&D necessary to build their products. Many of Oracle’s customers were betting on Oracle’s success and assuming that they would be getting better, more complete products in the future. I would argue that the vast majority of RDBMS innovation came about when there were many RDBMS vendors and Oracle was growing from $400 million/year to $1 billion/year. For the most part, support of the products was an afterthought. License sales were growing so fast that those fees far outpaced the support fees.

In the mainframe world support was a part of the license; if you didn’t pay support you lost the license. With Oracle and Microsoft you could license the product in perpetuity without ever buying support. At some point though, support became a more important aspect of the business. Unfortunately for companies like Oracle and Microsoft they can’t survive if they cut off the license revenue. Today Oracle collects billions in license fees for their database software but I don’t think anyone would argue that billions in new features and capabilities are delivered to the market. The market for licensed software is out of balance.

My statement regarding the product definition was, “…the proprietary bits are no longer the product.” This is true. Today, Red Hat could not sell the bits that make up Linux absent a support contract. Well, not to a very sophisticated customer anyway. Red Hat, instead, sells a support offering – timely, automatic updates, services around implementation, an infrastructure to assure availability of technical alerts and fixes, etc. The bits are freely available to anyone with the patience to assemble them themselves. Many do, but Red Hat is successful because of all the other companies who want a professionally managed operating system infrastructure.

Most of Oracle’s offerings include the same benefits and many of Microsoft’s stated directions imply the same thing. I believe it’s very easy for Oracle to justify their support fees to their customers, it’s the license fees that are, in my opinion, inappropriate.

The value of the bits in an open-source product is well established by the market: $0.00/bit. The value of support infrastructure and services is becoming well established by the market. Red Hat, Novell and others are monetizing that market.

Whenever I hear about hybrid open/closed-source business models I shudder. These business models exist only because someone still believes the proprietary bits contain value. It is the trusted assembly, servicing, support and delivery that contains the value. If you think your proprietary bits have value then be honest with your customers and ask them, “How much are you willing to pay me to write this code?” In today’s world, with a more ubiquitous skill set in operating systems, database systems, management systems and applications, I think it very difficult to justify collecting R&D dollars to build a future product. There’s just too much competition willing to collect on the back-end rather than the front-end.