From Y Combinator to Customer Traction and $50 Million in Financing: MemSQL CEO Eric Frenkiel (Part 2)

Sramana Mitra: How long did you stay at Facebook?

Eric Frenkiel: Not very long, it was for 10 months in total. It was an amazing opportunity. We ultimately left Facebook because we had a bigger opportunity that we wanted to build. That was MemSQL. We joined in 2010 and left in early 2011. That was a big decision because an extra two months would have meant a quarter of the shares, but it happened to coincide with YCombinator, which have fixed start dates. If you don’t start on the fixed date, you’d miss the class. In many ways, it was an easier decision because you just walk away. You look forward to the next venture.

Sramana Mitra: You took your idea and applied to YCombinator?

Eric Frenkiel: Yes.

Sramana Mitra: You had co-founders?

Eric Frenkiel: Absolutely. It’s a great story in terms of how the company was founded and how I met my co-founder Nikita. There are real joys of being in such a dense city of talent. There are a lot more opportunities. I met Nikita out of pure serendipity at Facebook on the very first day. We just happened to sit next to each other when we were going through orientation. Through that initial bonding we built a great friendship and worked together closely. When the time came for us to say, “Wow! This is an exciting thing,” it was a very natural discussion of leaving.

Sramana Mitra: What’s Nikita’s background?

Eric Frenkiel: Nikita is the technical visionary of MemSQL. He is a database engineer. I like to joke that he was put on earth to build databases. He’s been doing it his entire life from his initial undergrad in Computer Science to his Ph.D. and subsequently during his six or seven years at Microsoft. He was also recruited to Facebook to work on databases. He had vertical, deep expertise in databases.

The very reason why we started a database company was because he knows how to build them. It’s an easy thing to do when you have that expertise to leverage. We both were able to be great friends and when the idea came along, it was a partnership between Nikita and me to leave together.

Sramana Mitra: What was the idea?

Eric Frenkiel: The idea came about from seeing the major big data problems that Facebook was scaling years in advance. Being at the tip of the spear, Facebook was solving problems that companies are facing only today. Literally, you are seeing a four to five year lag between Facebook and what the industry is facing. Part of that problem was the fact that analytics was slow. Certain analytical work load at Facebook would take more than 24 hours to complete on a single query. Analytics latency was critical to solve.

Secondarily, the reason why we met on the same day was that Facebook has the concept of engineering boot camp. You join and you go to boot camp to get in shape. In Facebook sense, you do that to learn the systems of how to use and build a Facebook because all the tools had been de-normalized and reduced to its raw components. The system was at such a scale that it was the only way to achieve that scale. You spend eight weeks learning how to build Facebook because it’s not intuitive.

The interfaces are not intuitive. The data access patterns are exotic and unique. There’s a lot of glue between various systems to let it scale. We saw the friction. When you combine that non-intuitive interface with analytics latency, there seems to be a real avenue to solve that in the database chair. To make it fast, Facebook uses a lot of DRAM. The notion of putting data in memory was obvious, but to do that in an easy way took skills and expertise that Nikita had from his career. The basic concept was combining these two notions of easy-to-use interface and making it fast. We wanted to build an in-memory relational database. In a nutshell, MemSQL was designed to provide that Facebook-style scale-out and analytical speed to enterprise customers around the world.

This segment is part 2 in the series : From Y Combinator to Customer Traction and $50 Million in Financing: MemSQL CEO Eric Frenkiel
1 2 3 4 5 6