It’s not often that I am criticized for being technically imprecise. After all, I’m an open-source code monkey and proud of it. This time, the criticism came from a marketing guy, and even more surprisingly, he was right.

I was discussing the launch of an internal Google tool called ‘Google Mondrian’ with Lance Walter, who runs marketing at Pentaho, but is also technical enough to tell a big-endian half-adder from a Python cross-compiler if he were to trip over one in his front lawn. Google Mondrian is a web-based code-review tool written by Guido van Russum, the author of Python, and is apparently in widespread use internally within Google.

Our concern was that a high-profile announcement would somehow overshadow the Mondrian open-source OLAP project, but we fairly quickly concluded that this was unlikely. Supposing that this thing becomes as big for online code-reviews as, say, CruiseControl is for continuous integration, would that prevent people from finding and using Mondrian? Probably not. And the nice thing about the launch was the slashdottage from people leaping to our defense.

But, I concluded, “we’d probably better use the phrase ‘Mondrian OLAP Server’ so that no one gets confused”.

Lance wasn’t having that. “I swear I heard you at one point say that it was technically innaccurate/misleading to call Mondrian an OLAP Server. You designed Mondrian to be lightweight, so it could just as easily be embedded in an application as run as a server in a three-tier architecture.”

He was right, of course. I was not being consistent. Mondrian is variously described as “an OLAP (online analytical processing) database written in Java”, “an OLAP server written in Java” and “an Open Source OLAP (online analytical processing) server, written in the Java programming language”. (I didn’t write the last one.) “The Mondrian OLAP engine” is a more accurate term, because “engine” has been used to describe embeddable relational databases like Apache Derby and the Microsoft JET Engine. It’s easy to accidentally use the phrase “OLAP Server” because the first commercial OLAP engines all had I’m-the-center-of-the-Universe architectures – the logical thing to do if you’re trying to shift $1M of software, because people will pay a lot more for a server than for a library – and so that’s become the standard term.

It’s a Faustian dilemma which often crops up in technical marketing. Do you describe what your product actually is and does, or do you tailor your description to what you think your audience is looking for, even if that is a little inaccurate? Incidentally, I make no apologies for ‘marketing’ Mondrian. Even if I believed that no one should ever pay another cent for software (and I don’t – I’m not that much of an idealist), I would still have a duty to describe and publicize my project so that its target audience can find it. I’d rather that people find Mondrian by googling [see note below] “olap server”, then read the fine print and discover, “Wow, it’s embeddable too!”, than not to find it at all.

So, I suspect that Mondrian will remain the “Mondrian OLAP Server” or, to its friends, just “Mondrian”.

Note to Google’s legal department: I know you don’t like me to use ‘google’ as a verb. It must be horrible to be so successful that you are part of the English language, as in ‘I xeroxed a polariod of my jeep’s collision with a dumpster’. I really feel for you. But since we’re now sharing the term ‘mondrian’, I figure you owe me one.