On Wednesday, November 7, 2012, I attended November Tech Talk: The State of NYC Investing hosted by ZocDocs. Two presenters gave short demos: Matt Witheiler, Principal at Flybridge Capital Partners and two members of the Thrillist Media Group: Jay Chinthrajah and Ken Peltzer.
Matt Witheiler introduced himself as a Principal at Flybridge Capital Partners. He started his presentation by reasoning his move from Boston to New York City. Of course, he didn’t bat an eye comparing the New York Yankees to the Boston Red Sox, drawing a lot of “boos” from the audience.
Witheiler went into explaining what a “Bridge and Tunnel” investor means. “I was a bridge and tunnel investor for two years,” Witheiler said. “That means that investors — like me — parachute into New York and then go back to their homes.” For two years Witheiler practiced his bridge and tunnel investing for two years because of New York City’s larger number of deals for both ends. In the Spring of 2012, however, Withlier and his firm sat down to review data and committed his move to New York City.
For Witheiler, moving to New York City came down to four reasons: 1) New York City has a strong and welcoming entrepreneurial community. 2) Large West Coast-based tech companies are setting up shop in NYC—Twitter, Google, eBay, Microsoft all have engineering teams here developing tech in NYC. 3) There are huge emerging leaders in many categories in NY — covers a wide spectrum. 4) Broad macro support for the entrepreneurial community in NYC.
There are risks that come with appropriating a chance to move to a different city as an investor. “There is a mismatch in the number of seed funds and the number of series A and funds that are not sustainable,” Witheiler said. There are Seed/Angel funds with fulltime people in New York City that have 15 to 20 startups in their portfolio. There are also seed to scale funds with fulltime people in New York City. The majority of Seed/Angel funds want to be funded and moved to the next level.
The New York City exit environment gets knocked, which is another risk that Witheiler explained he had to consider when moving to NYC. However, Witheiler believes that “in the next 12-18 months, some of the company exits will be more than a billion dollars.” This will help NYC’s exit environment and attract investors. The increase in attention will increase valuation. The downside to this, at least, to Witheiler is that “if prices continue to rise, it risks losing investors and companies lose funding…. You either take the money or take money elsewhere. New York isn’t a bargain anymore,” he said.
When asked why engineers move from Boston to other cities, Witheiler said that Boston doesn’t have a strong community for entrepreneurs in the tech field so developers leave for other cities.
Thrillist Media Group was next to present. Jay Chinthrajah and Ken Peltzer showcased their creation, Zuul, “the gatekeeper to the Thrillist User-verse.”
There are two teams currently within the Thrillist Media Group: Thrillist and JackThreads — content and commerce, listed respectively. A problem they faced was that both systems are different to maintain; that is, it is hard to analyze performance analytics simultaneously. “How do you store and perform analytics across two systems?” Chinthrajah asked. Their solution was to create Zuul, a unified user system.
Zuul, according to Chinthrajah and Peltzer, is extensible and future-proof. It is fast and can handle traffic without disrupting page view and provides reliability, allowing a hands-off approach.
Phase 1 in developing Zuul came in three stages: describing the problem, asking, “What are we going to build?” as it has to be responsible and easy enough to plug in and go, and deciding on and learning new technology.
Zuul was written in Ruby using the Sinatra framework. Nginx and Unicorn are used for web servers and Memcached for caching GET requests. MySQL is its primary datastore and MongoDB is used for user searching, aggregating and logs. Redis is used for queue storage and background apps. Reque is used for background job processing and GOD is used for system monitoring.
Zuul’s phase 2 was building to alpha. “This was a unique way for Thrillist because we got to build what we wanted to,” Peltzer said. It took three weeks of isolated development for the team to complete the build. “There were mountains of code and epic business rule battles,” Peltzer said. “We learned here to not to talk to anyone. It was the most productive way to approach this project.”
Phase 3 was to integrate content and commerce. Chinthrajah and Peltzer brought the Thrillist and JackThreads team together to replace old user system guts with API calls.
The final phase was the testing phase where Chinthrajah and Peltzer brough on a dedicated QA team and tested Zuul using real, live traffic and “beat the living hell out of it for a long time.”
Zuul’s launch was successful, recording “less than 40ms response time at peak load.” It is expandable—every component can be separated from the core and upgraded if needed. Zuul was tested up to 10x peak load and “prepared for the future.”
“We learned to be wary of service consumers and to ensure data integrity if you’re converting. You should also know the ins-and-outs of your entire stack,” Chinthrajah said.