I received an email from Tarus Balog, CEO of OpenNMS Group, on Friday, taking issue with the language I had used to describe two open source vendors (and I use that term deliberately).

Essentially Tarus objected to me using the term “open source vendor” to describe two companies with Open Core licensing strategies. His email raises a valid point about how we determine which companies are considered “open source vendors” and I wanted to use the opportunity to outline the rules I use to make that decision.

As a technical snafu at our end had prevented Tarus from leaving a comment on the blog I hope he won’t mind me using his words to explain the issue he raised.

He wrote:

“You posted:

‘Meanwhile two of the biggest open source vendors, Hyperic and JasperSoft announced a partnership which sees JasperServer Professional Edition is being embedded in the new Hyperic Operations IQ, a new business intelligence platform for IT and web operations teams.’

To paraphrase Vizzini from the Princess Bride, when you say ‘open source’ I do not think those words mean what you think they mean.

Hyperic IQ is commercial software. There is no open version, and it’s published under a closed license (at least from what I can gather). Both Hyperic and JasperSoft are ‘open core’ companies - commercial
software companies who use open source to market their software. So it seems a little ironic that you would preface this news, where they don’t even make the pretense of having a ‘community’ version, by
calling these companies ‘open source vendors’.”

This is an issue that has raised its head before. I wrote last May about what I perceived to be evidence of tension between commercial open source vendors and elements of the open source user community over the use of proprietary licenses for open source code, or extensions to open source code.

Tarus has been writing about the subject quite in recent months and in December wrote about what he called “The War for Open Source“and the use of the term “open source” by vendors that are using Open Core licensing strategies.

Tarus noted that he used the Open Source Definition to determine what he saw as open source. Which is fine, apart from the fact that the OSD only talks about the software, not the vendor business strategy used to commercialise that software.

Additionally, as previously noted, the OSD only defines the license used to distribute the software, not the method used to develop it. There is nothing in the OSD that mandates a community (although its terms do encourage community development).

Tarus later followed that up with the suggestion that people should use “The CentOS Test” to decide whether a product is open source or not:

“When thinking about a purchase of the paid or “enterprise” version of something labeled open source software, ask yourself “does it pass the CentOS test?” Examine the license to see if it would be possible for you to take the source code, compile it and distribute it. If you can, I claim it is likely the software is truly open. If not, then you are looking at commercial software, with all of its limitations.”

I believe this test is a pretty efficient method for determining whether a product is open source or not. However it does not help decide whether the vendor could be referred to as an “open source vendor” or not if that vendor also has products that don’t pass the test.

Using the vendors Tarus referred to in his email as an example, Hyperic HQ Open Source would pass the test. Hyperic HQ Enterprise and Hyperic Operations IQ would not.

Similarly, JasperServer Community Edition would pass the test. JasperServer Professional would not.

So would it be correct to call Hyperic and JasperSoft “open source vendors”? In my response to Tarus I noted that:

“Given that the main products of both companies are based on an open source core I think a lot of people would consider them to be ‘open source vendors’ although given there is no official definition for what makes a company (as opposed to a product) ‘open source’ that is a matter of personal opinion.”

Based on Tarus’ comments I did make a quick amendment to the post he mentioned and I will try to avoid generalisations, but I also don’t want to get into a position where we have to preface every mention of a company with an explanation of its business model.

What makes a company an “open source vendor” is something I have to consider often when writing and researching reports for The 451 Group. It was particularly relevant for the Open Source is Not a Business Model report, and is also top of mind now as I am preparing to write a report about venture capital investment in open source vendors. Clearly with reports such as these we have to draw a line somewhere.

In writing these reports I made the decision that for me an “open source vendor” is one that is reliant on open source software in order for its business strategy to function. By that definition we would include open source purists such as OpenNMS but also the likes of Hyperic and JasperSoft, which would have very different commercial business strategies were their products not based on open source.

This is also the reason our coverage includes hardware vendors that are entirely reliant on open source software within their products, such as Linux-focused server makers like Penguin Computing and Collax. For the same reason we also include software vendors that build traditionally licensed products on open-source-developed code, such as Greenplum and Bluenog.

The definition is far from perfect, and it is sometimes difficult to decide whether a vendor should be included or not (for example Kickfire, which makes database appliances which include the commercial MySQL license, but would not have been able to have built the business or be targeting the MySQL user base were it not for the open source version).

The definition I use is not the definition of “open source vendor” but it is my definition, and it is the one I will continue to use until there is some universally approved definition.