Paul Dix is holding on to his Williamsburg apartment. The CEO and cofounder of InfluxDB has strong personal ties to Brooklyn and it’s unlikely that he’ll totally vacate the place any time soon. However, his investors wanted his time series database company on the West Coast, where they believed it could find the right talent to grow.
That’s why, Dix explained, he and his cofounder shifted their HQ from New York City to San Francisco six weeks ago. Dix has apartments in both cities now.
"I love the New York City tech scene, but anyone who tells themselves it won't be easier to raise money out there is crazy."
“The vast majority of Valley-based investors said the company has to be in the Valley,” Dix told us, saying this reality and the pool of experienced talent out there is still really important, especially in a space that’s very B2B oriented. He added, “I love the New York City tech scene, but anyone who tells themselves it won’t be easier to raise money out there is crazy.”
We covered InfluxDB’s $9.1 million Series A round in a Startup Roundup last December. TechCrunch did a good explainer on the company’s origin story and what it’s doing. In short, Dix brought his cofounder, Todd Persen, onto a freelance project that they both did for Thomson Reuters. They liked working together and they both wanted to start something, so they came up with the idea of building a SaaS product that would help companies with metrics and spotting aberrations.
The only problem was that lots and lots of companies are building the same thing. However, Dix and Persen realized that they had built a backend that all these other companies had to have built, too: a time series database. The crux of a time series database is that it’s good at logging data over time. Obvious use case: stock prices. The financial industry had been into time series data for a long time, Dix told us, but these days everyone wants time series data. Google Analytics may be the most famous application here, and every time someone puts out a product using time series data, they did all the work to build their own time series database.
Suddenly, Dix and Persen lost interest in their time series data product and threw all their effort into the database itself.
The idea was that they could build that open source and, if they did a good job, then devs would pitch in, spread the word and investors would back them so that they could build up enough of a user base that a scaleable business could arise from the project. Dix said dozens of devs that have no ties to his company all over the world have given talks on InfluxDB at conferences and meetups over the last year, simply because they like the database.
Which raises the question: How do you build a profitable company by giving your core product away for free?
Dix is quick to point out that several companies have already done it. Datastax is built on Apache Cassandra. MongoDB, Inc. is built on MongoDB. Hortonworks had a $100 million IPO in December, and it’s built around Hadoop.
"The game plan for an open source company is for the first several years you get the user base as big as you can."
The way you do it, Dix explained, is first and foremost by building up a really great open-source product that developers are excited about. Then you sell enterprise licenses, support services, training and professional services. On top of that, you build closed source add-ons for the core open source product that companies can buy if they want. Dix said they’ll release some of these, as a first pass, this year. Though their emphasis is still squarely on InfluxDB.
“The game plan for an open source company is for the first several years you get the user base as big as you can,” Dix told us. They haven’t done any traditional marketing yet, though that is coming. One nice thing about starting open source, Dix explained, is that people will let you come and talk at meetups and conferences where a closed-source company wouldn’t be permitted. Because you’re giving something away, and contributing, people are open to hearing from you. He said he probably gave forty talks around New York City in the company’s first year.
InfluxDB is winning adherents for a few reasons, he argues. It uses queries that are similar to SQL, so that feels familiar and it has downsampling and averaging built into the product. This is important because if you query a time series database, you’re looking at an enormous amount of data. For the computations not to be painfully slow, sampling is key.
Additionally, because delivering similar services are all built into SaaS products, InfluxDB can democratize some of these use cases and let good devs in startups use InfluxDB to build out exactly the tool for their needs, rather than paying for a third-party product they may not have the cash for. For Philadelphia startup Squareknot, using Backbone.js made it feel like they instantly expanded their dev team.
InfluxDB is currently a team of 10 with two contractors. While it has a San Francisco HQ, it also recently set up a Denver outpost because they found some good developers there. Dix splits his time between the two coasts.