Before joining Terracotta I had spent close to decade at Network Appliance where we perfected a business model built on extremely proprietary technology - we referred to our products as tin-wrapped software. So when Terracotta CTO Ari Zilka proposed an open source model based on the observation that we could align better with our OSS customers and partners - I understandably balked at the prospect. Terracotta clustering technology has garnered acclaim for being innovative, and the thought of opening it up and commoditizing it seemed counter-intuitive and counterproductive to our goal of building a healthy business.

It's not that I couldn't see the advantages of going open source. Clearly, if we made Terracotta code open source, the OSS community would be able to use it and extend it freely and overcome many of the concerns around trying new proprietary software. This would be much easier than trying to win over hearts and minds one encounter at a time. Over time the community would also be able to help us improve our technology and provide valuable input on direction.

My primary doubt was identical to those of all my colleagues who have had encounters with OSS. How do you build a sustainable business if you make your software freely available to your customers and competitors? The quick answer that I received from other OSS-based companies - most of whom were not yet profitable - was the standard training, support, and services! While that made some sense, I had trouble visualizing it.

Surprisingly, I had to return to my experiences in the proprietary world to convince myself that we could build a healthy business around an Open Source Terracotta. About 10 years ago at NetApp we had a product called NetCache that started life as the proprietary sister of Squid - an open source proxy server. Initially, we tried to compete with Squid and were strongly rebuffed by developers who liked its flexibility and the fact that it was open. In the end we realized that we would have to complement Squid. We, in fact, encouraged developers to use Squid and built a business model around strengthening large-scale Squid operations. We targeted very large users of Squid and sold them on the benefits of enterprise-class support, advanced monitoring, load balancing, and other value that operators can clearly perceive.

Reflecting on this experience, the Terracotta team realized that we should never charge developers. We have to make everything that the developer needs to go into production open to them. Terracotta's revenues will come from IT operations. We believe that the operators of large deployments of Terracotta will need and want to pay for enterprise-class support and services, and down the road will want us to create new products that will help simplify large-scale operations.

Of course, this approach means that the majority of Terracotta users will not have the scale to ever have to pay us. We accept that. We believe that without open sourcing we will not have reached most of those users. Indeed most of those users do not cluster their applications today because of the complexity and cost of doing so. In this scenario they are able to benefit from using Terracotta and we benefit from their guidance and contribution.

So I come full circle to where we are today. My discussions with and observations of successful OSS-based companies has led to my belief that OSS cannot be a marketing exercise - the effort that we have put in over the last six months in preparation for our community has changed the company's focal point. For Terracotta to be successful it is clear to us that we have to make building a healthy and thriving community our primary goal.