What do we know about open source pricing?
Is there a formula for what open source software should cost? Maybe not, but open source vendors should endeavor to make pricing transparent
Follow @infoworldBy Mark de Visser
Pricing for open source software products remains a hotly debated topic of interest. Dave Rosenberg argues in a recent post on Negative Approach that there is consensus that the price of open source software should be 10 to 20 percent of the price of equivalent proprietary software. He does not attribute that to anyone, and in our echo chamber of technology bloggers and tweeters, such a statement will quickly come to be seen as a long-established fact.
Initially, my intuition told me that Rosenberg's thesis was wrong, but I didn't want to dismiss it without doing my own research. I looked into it a bit more and uncovered some interesting facts that argue both for and against the 10-to-20 percent range put forward by Rosenberg.
[ For a recent sample of InfoWorld's in-depth open source coverage, check out specialty Linuxes to the rescue. ]
My first experience with open source pricing was in 2002, when I was at Red Hat. I was part of team that had to set pricing for Enterprise Linux (or Red Hat Advanced Server, as it was called in those early days). We consciously decided to price it about 20 percent above Microsoft's equivalent Windows servers, which we viewed as our major competition. Both Red Hat and Microsoft aimed to convert Unix users to the Intel platform, and we felt lower pricing would harm our credibility as a competitor.
Many will argue that Enterprise Linux succeeded because the cost of the stack of Intel hardware plus Linux OS was lower than the equivalent Sparc/Solaris stack, but even that aggregate cost benefit was not 80 percent or more -- nor did it need to be. Today the prices at OS vendors Red Hat, Sun, and Microsoft remain at about the same levels. Open source or not makes little difference; Red Hat and Sun use other arguments in their competition with the proprietary alternative from Microsoft. Rosenberg's thesis does not hold here.
I also looked at the 12 companies that signed the Collaborative Software Initiative's Open Source Letter to Obama and was surprised to see that only 3 of these 12 disclose their pricing on their Web sites. Compiere prices its professional ERP service at $50 per user per year, a lot less and a lot simpler that the $1,500 to $5,000 per user for Oracle's basic ERP product. With that math, Compiere goes beyond supporting Rosenberg's thesis, falling way below his suggested range.
Ingres calculates its TCO over three years at $575,000 versus $960,000 for Microsoft SQL Server; $2.5 million for IBM DB2; and $3.7 million for Oracle DB, a discount ranging from 40 up to 84 percent against proprietary products. OK, Dave, I'll give you that one too.
