This is an interesting story of how an open source software company built around Cassandra was incubated by RackSpace and has grown to $5 million in revenue. Founded by engineers Jonathan Ellis and Matt Pfeil, the interview traces not only the successes of their journey but also the mistakes they made in structuring their funding rounds.
Sramana Mitra: Jonathan and Matt, let’s start with both of your backgrounds. Where you were born? Where did you grow up? How did you get together?
Jonathan Ellis: I grew up in New Jersey. I met Matt after I moved to Texas to work for Rackspace. Rackspace hired me to build a scalable database for their internal infrastructure as they started to compete more with companies like Amazon, Google, and the Cloud. In late 2008, I started working on Cassandra. I met Matt Pfeil shortly afterwards as he led the group that was going to be deploying Cassandra internally at Rackspace.
My background in big data started a little before that when I built a storage engine for the backup provider Mozi. I built an object store that let people back up their files to Mozi’s data centers. This was a petabyte scale object store, so it scaled out quite nicely, but it’s not a database. It was not designed for low latency query and retrieval. It was specifically designed for storing large blogs for a backup engine. So I was working on a general purpose database that could be used for driving applications. To scale was a logical next step for me and that’s why I was interested in working with Rackspace on that.
Sramana Mitra: And Matt, what about you? Where are you from? What’s your background?
Matt Pfeil: I went to school in Virginia Tech. I grew up in Europe and the Middle East because my dad was in the Navy. Right after Virginia Tech, with a Computer Science degree in hand, I worked with a little startup that was in the same town. It was an e-mail hosting company for small businesses, a part of which was actually acquired by Rackspace. I was working on its backend and had a lot of experience storing large numbers of very small files. We were doing things like hacking relational databases just to make this big data problem work on a relatively large scale.
After Rackspace bought us, I moved to Texas at their request. I was working on building out shared infrastructure for the various groups at the Rackspace Club, so they could focus on building customer-based features as opposed to scaling their data source. And the funny part of that story with Jonathan is that he told me he was going to leave. I spent a lot of my time recruiting at that point. Whenever I took him out to lunch to try to convince him not to leave, I remember asking, “Who the hell is going to do the business side for you?” And he said, “Well, you can do it.” That was not an argument that I won on that day.
Sramana Mitra: And did you guys know while you were still working together at Rackspace what the startup was going to be about?
Jonathan Ellis: Not really, because we didn’t know that we wanted to build a product company. We didn’t know exactly what that looked like, so started up with the typical open source playbook of services and support. We built a management tool as our first step towards a product-based business. But it was more of ‘we’ll know it when we see it’. Building an enterprise product around Cassandra was the next step we were looking for. We didn’t know that’s what we’re going to build when we started it.
Matt Pfeil: One of the things that became apparent very early on was when we started offering service SLA support for production clusters. First of all, it was really nice that we had customers on day one. It’s because of how successful Cassandra had been while we were working at Rackspace. That was a huge benefit but then as we started offering services, no one really wanted to get a pager at 3:00 in the morning. So, one of the first products we started building was OpsCenter, our management offering. Rather than having a human-based reactor support, we tried to build tools that would help people do that in an automated way. It’s really nice to build things that address their pain.