SM: What was your ultimate design goal?
MM: The first problem we aimed to solve was how to make a re-transmission machine, a control system that retransmits data at a rate that matches the channel capacity for any packet loss rate over the channel and any delay. That was our design goal.
My contribution was finding a model that seemed to make sense. We implemented accordingly. After we got the essence of it, we began to discover various corner cases where it would begin to break down. It took us a good part of five months or so to get the basic retransmit algorithm worked out. Not only how it would work, but also how to implement it in software. That final piece is to Serban’s credit. The data structures do not slow down as retransmissions accrue.
SM: Good innovation.
MM: The other great thing is that because we were precise about what we had decided to make, we had an easy way to measure it. We had our dummy net which introduced LAN loss. We had our sender and receiver which allowed us to set up experiments where we changed the LAN packet loss. We looked at the rates we were able to achieve, and we had counters everywhere.
SM: You built the algorithm and the simulator in the 12 months prior to launching the company?
MM: Yes. That was all I was doing. I did not have any other job. I was very excited about it so I did it non-stop. I give Serban full credit for his software engineering. The model is correct and has proven itself. The software engineering is what makes it run so well. Serban designed all of the algorithms we used, and the data structures, and he implemented it. He is an outstanding engineer.
We did a second generation on the bandwidth control, which was very interesting. That was also my idea to pursue it. Once we got going it turned into the Socratic method with plenty of back and forth. We worked very well together and have worked with each other so long that this was just a conversation we had all the time, every day. We would challenge why we were doing something a particular way and why it was different from TCP. Sometimes you do things but you are not sure why you are doing them. You have to constantly reconcile.
SM: What is important is that you found your business soulmate.
MM: Yes. We had worked together a lot before. We did a lot of work at the prior company, where we stayed until midnight. Neither of us started Aspera for the sake of having a huge company. We were both extremely interested in this area, and we wanted to control our own destiny.
SM: How did you structure the company?
MM: It is very simple with a near 50/50 split. Another very easy decision was regarding our manner of funding. We did not want outside control.
SM: Did you give your father any equity?
MM: He has a bit, as he should. If you looked at our equity structure you would find it matches our company history nearly perfectly. About 25% of our company is owned by our employees. I think we have 400 customers now.
SM: Who are your customers nowadays?
MM: They are customers who are largely involved in media. The big studios are the ones you think of, but there are plenty of folks down the chain now. There are video-on-demand players and online aggregators. We have DirectTV, Comcast, EchoStar, TimeWarner and others on the VoD angle. On the aggregators there is Amazon for the video store, Microsoft for the Xbox content and others. Cloud computing, Amazon and IBM are all of interest because what they are being asked to do has some digital media emphasis plus a large amount of other data.
SM: Your customers are mostly large companies, so I am assuming you have a direct salesforce. Is that correct?
MM: Yes. We are almost exclusively a direct salesforce, although we are working on building up a channel. Today we have seven people in sales. We have one person in each region. We have a guy in Europe, someone on the East Coast, a person in the center [of the US], a guy out West, a couple of guys on top and a new business development guy. They just spread out. The company has changed, in the business sense, like night and day in the past two years. It has doubled in the number of employees each year we have started. Because we never did the VC thing we never had a big glut of hiring. The hiring just follows the business.
SM: That is the right way to do it whether you have VC or not.
MM: We certainly tried.